{"id":1403,"date":"2023-08-29T21:59:36","date_gmt":"2023-08-29T19:59:36","guid":{"rendered":"https:\/\/evaltep.xcreative.cz\/uncategorized\/diskuze-metodickych-postupu-pri-evaluaci-dopadu-aktivni-politiky-zamestnanosti-cr-na-datech-okprace\/"},"modified":"2023-09-15T15:43:19","modified_gmt":"2023-09-15T13:43:19","slug":"discussion-of-methods-and-approaches-towards-the-impact-evaluation-of-czech-active-labor-market-policy-using-okprace-data","status":"publish","type":"post","link":"https:\/\/evaltep.xcreative.cz\/en\/articles\/discussion-of-methods-and-approaches-towards-the-impact-evaluation-of-czech-active-labor-market-policy-using-okprace-data\/","title":{"rendered":"Discussion of Methods and Approaches towards the\u00a0Impact Evaluation of Czech Active Labor Market Policy Using \u201eOKpr\u00e1ce\u201c Data"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\"><strong>Abstrakt<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Evaluace dopadu je spojena s&nbsp;mnoha metodologick\u00fdmi volbami ze strany v\u00fdzkumn\u00edka. V\u00fdsledky evaluace jsou pak zna\u010dn\u011b z\u00e1visl\u00e9 na definici konkr\u00e9tn\u00edho evalua\u010dn\u00edho modelu. C\u00edlem tohoto \u010dl\u00e1nku je diskutovat z\u00e1kladn\u00ed oblasti, principy a postupy hodnocen\u00ed v\u00fdsledk\u016f a dopad\u016f program\u016f aktivn\u00ed politiky zam\u011bstnanosti (APZ) s&nbsp;vyu\u017eit\u00edm administrativn\u00edch dat syst\u00e9mu OKpr\u00e1ce. Metodologick\u00fdm postupem p\u0159itom m\u00e1me na mysli sekvenci voleb a d\u00edl\u010d\u00edch procedur evaluace, vedouc\u00ed k validn\u00edm v\u00fdsledk\u016fm. \u010cl\u00e1nek se zab\u00fdv\u00e1 prezentac\u00ed postup\u016f, kter\u00e9 jsou vhodn\u00e9 pro individualizovan\u00e1 data OKpr\u00e1ce a opom\u00edj\u00ed postupy, kter\u00e9 nelze na t\u011bchto datech uspokojuj\u00edc\u00edm zp\u016fsobem vyu\u017e\u00edt. \u010cl\u00e1nek uplat\u0148uje perspektivu tzv. \u010d\u00e1ste\u010dn\u00e9ho ekvilibria, kdy hodnot\u00ed dopady program\u016f na jejich \u00fa\u010dastn\u00edky a&nbsp;nikoli na trh pr\u00e1ce jako celek. J\u00e1drem je diskuze postup\u016f s&nbsp;ohledem na&nbsp;intern\u00ed validitu nam\u011b\u0159en\u00e9ho dopadu. Jako prost\u0159edek pro dodate\u010dn\u00e9 vybalancov\u00e1n\u00ed p\u0159edprogramov\u00fdch rozd\u00edl\u016f mezi skupinou intervence a&nbsp;kontroln\u00ed skupinou diskutuje proces p\u00e1rov\u00e1n\u00ed, p\u0159i\u010dem\u017e zmi\u0148uje d\u016fle\u017eitost zahrnut\u00ed faktor\u016f p\u0159edprogramov\u00e1 perspektiva a historie nezam\u011bstnan\u00fdch na trhu pr\u00e1ce a f\u00e1ze hospod\u00e1\u0159sk\u00e9ho cyklu. \u010cl\u00e1nek uzav\u00edr\u00e1, \u017ee v\u00fdb\u011br konkr\u00e9tn\u00ed techniky anal\u00fdzy dat z\u00e1vis\u00ed na tom, zda chce evalu\u00e1tor zjistit velikost absolutn\u00edho nebo relativn\u00edho rizika, jeho pr\u016fb\u011bh v&nbsp;\u010dase<br>anebo&nbsp;pouze p\u0159\u00edtomnost nezam\u011bstnan\u00e9ho v&nbsp;registru.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Abstract<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Realization of impact evaluation brings possibility of various methodological choices that are limited to distinct assumptions concerning the context of the study, nature of the data etc. Results of the impact evaluation are then dependent considerably on definition of particular evaluation model. Aim of this article is to discuss main areas, principles and approaches to&nbsp;evaluation of Czech active labor policy programs (ALMP) outcomes and impacts, using administrative data of OKpr\u00e1ce. The article presents approaches that are suitable for this kind of data and is based on partial equilibrium perspective, thus omitting impacts on labor market as whole. The core of the article is discussion of diverse methodical steps in context of internal validity of measured impact. The matching is discussed as&nbsp;a&nbsp;method of balancing pre-program differences between program participants and control group. Inclusion of factors of labor market perspective and history, as well as phase of economic cycle, during this process is of&nbsp;high importance. Article concludes that choice of particular statistical method of data analysis hinges on whether evaluator wants to find absolute or relative risks of leaving unemployment, their progress in time or&nbsp;merely the presence of unemployed in register.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Kl\u00ed\u010dov\u00e1 slova<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Aktivn\u00ed politika zam\u011bstnanosti, dopadov\u00e1 evaluace, kvazi-experiment\u00e1ln\u00ed design, OKpr\u00e1ce<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Keywords<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Active labor market policy, impact evaluation, quasi-experimental design, OKpr\u00e1ce<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>\u00davod<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Tato studie se zab\u00fdv\u00e1 metodologick\u00fdm postupem programov\u00e9ho hodnocen\u00ed program\u016f<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn1\"><sup>[1]<\/sup><\/a>&nbsp;aktivn\u00ed politiky zam\u011bstnanosti (APZ) na individualizovan\u00fdch datech OKpr\u00e1ce. Zab\u00fdv\u00e1me se evaluac\u00ed v&nbsp;perspektiv\u011b tzv. \u010d\u00e1ste\u010dn\u00e9ho ekvilibria (mikro-ekonomickou evaluac\u00ed), tedy efektem intervence na&nbsp;jej\u00ed \u00fa\u010dastn\u00edky, nikoli na trh pr\u00e1ce jako celek. Metodologick\u00fdm postupem mysl\u00edme sekvenci voleb a d\u00edl\u010d\u00edch procedur evaluace, vedouc\u00ed k validn\u00edm v\u00fdsledk\u016fm. Vych\u00e1z\u00edme z&nbsp;p\u0159edpokladu, \u017ee zde rozpracovan\u00e9 kvazi-experiment\u00e1ln\u00ed postupy mohou b\u00fdt vhodnou \u201edruhou nejlep\u0161\u00ed volbou\u201c (v\u00fdsledkov\u011b p\u0159im\u011b\u0159en\u011b srovnatelnou s&nbsp;proveden\u00edm experimentu)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn2\"><sup>[2]<\/sup><\/a>. Z\u00e1rove\u0148 v&nbsp;p\u0159\u00edpad\u011b dat OKpr\u00e1ce lze v&nbsp;t\u011bchto kvazi-experiment\u00e1ln\u00edch postupech nal\u00e9zt nesporn\u00e9 v\u00fdhody:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>zna\u010dn\u00fd rozsah sledovan\u00fdch dat umo\u017e\u0148uje bohatost zji\u0161t\u011bn\u00ed v&nbsp;oblasti dopad\u016f program\u016f na r\u016fzn\u00e9 skupiny nezam\u011bstnan\u00fdch za r\u016fzn\u00fdch podm\u00ednek,<\/li>\n\n\n\n<li>k&nbsp;dispozici je \u00fapln\u00e1 populace nezam\u011bstnan\u00fdch,<\/li>\n\n\n\n<li>evaluace je mnohem levn\u011bj\u0161\u00ed ve&nbsp;srovn\u00e1n\u00ed s&nbsp;prav\u00fdm experimentem.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Evaluace dopadu je zalo\u017eena na \u0159ad\u011b p\u0159edpoklad\u016f spojen\u00fdch s jednotliv\u00fdmi postupy a na mnoha metodologick\u00fdch volb\u00e1ch ze strany v\u00fdzkumn\u00edka. S&nbsp;ur\u010ditou nads\u00e1zkou lze \u0159\u00edci, \u017ee samotn\u00fd evalua\u010dn\u00ed postup je u \u0159ady evaluac\u00ed druhou pomyslnou \u010dernou sk\u0159\u00ed\u0148kou s&nbsp;\u0159adou nevyjasn\u011bn\u00fdch evalua\u010dn\u00edch m\u00edst (prvn\u00ed \u010dernou sk\u0159\u00ed\u0148kou jsou nezn\u00e1m\u00e9 podm\u00ednky p\u016fsob\u00edc\u00ed uvnit\u0159 programu). V\u00fdsledky evaluace jsou pak zna\u010dn\u011b z\u00e1visl\u00e9 na definici konkr\u00e9tn\u00edho modelu (Betcherman, Olivas a Dar 2004).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">C\u00edlem studie je definovat z\u00e1kladn\u00ed oblasti, principy a postupy hodnocen\u00ed v\u00fdsledk\u016f a dopad\u016f program\u016f APZ s&nbsp;vyu\u017eit\u00edm administrativn\u00edch dat syst\u00e9mu OKpr\u00e1ce. D\u016fraz je kladen zejm\u00e9na na zhodnocen\u00ed postup\u016f tak, aby&nbsp;v\u00fdsledkem jejich pou\u017eit\u00ed byl intern\u011b validn\u00ed indik\u00e1tor dopadu programu, tedy takov\u00fd indik\u00e1tor, je\u017e by se maxim\u00e1ln\u011b bl\u00ed\u017eil skute\u010dn\u00e9mu kauz\u00e1ln\u00edmu efektu programu na v\u00fdsledek. C\u00edlem v\u0161ak nen\u00ed zpracov\u00e1n\u00ed metodiky evaluace dopadu, n\u00fdbr\u017e diskuze z\u00e1kladn\u00edch oblast\u00ed, princip\u016f a&nbsp;postup\u016f, na kter\u00fdch bude budouc\u00ed metodika zalo\u017eena. Studie je ur\u010dena zejm\u00e9na zadavatel\u016fm dopadov\u00fdch evaluac\u00ed, analytik\u016fm \u00daP a ostatn\u00edm realiz\u00e1tor\u016fm dopadov\u00fdch evaluac\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Syst\u00e9m OKpr\u00e1ce je vyu\u017e\u00edv\u00e1n ke spr\u00e1v\u011b administrativn\u00edch dat vkl\u00e1dan\u00fdch pracovn\u00edky jednotliv\u00fdch pobo\u010dek \u00da\u0159adu pr\u00e1ce \u010cR. Hlavn\u00edm \u00fa\u010delem tohoto syst\u00e9mu je p\u0159edev\u0161\u00edm v\u00e9st evidenci o jednotliv\u00fdch nezam\u011bstnan\u00fdch, n\u00e1roc\u00edch na d\u00e1vky soci\u00e1ln\u00edho zabezpe\u010den\u00ed \u010di spravovat informace o APZ. Z\u00e1rove\u0148 je ov\u0161em vhodn\u00fdm zdrojem dat pro sekund\u00e1rn\u00ed anal\u00fdzu APZ<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn3\"><sup>[3]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Pro hled\u00e1n\u00ed vhodn\u00e9ho postupu anal\u00fdzy vyu\u017e\u00edv\u00e1me jako podklad p\u0159edev\u0161\u00edm zahrani\u010dn\u00ed metodologickou literaturu. Rozhoduj\u00edc\u00ed pro posouzen\u00ed jednotliv\u00fdch dopad\u016f je ale p\u0159edev\u0161\u00edm intepretace spole\u010densk\u00e9 hodnoty uveden\u00e9ho postupu a dosa\u017een\u00e9ho dopadu ze strany u\u017eivatel\u016f. Studie je koncipov\u00e1na tak, aby vyhovovala zp\u016fsobu definice dat v&nbsp;syst\u00e9mu OKpr\u00e1ce \u010di podobn\u00e9m syst\u00e9mu, nebo\u0165 je pravd\u011bpodobn\u00e9, \u017ee z\u00e1kladn\u00ed po\u017eadavky na funkcionalitu a formu syst\u00e9mu administrace d\u00e1vek budou i p\u0159i p\u0159\u00edpadn\u00e9 budouc\u00ed zm\u011bn\u011b dodavatele syst\u00e9mu (po vyhl\u00e1\u0161en\u00ed v\u00fdb\u011brov\u00e9ho \u0159\u00edzen\u00ed) zachov\u00e1ny. Tato studie se tedy zab\u00fdv\u00e1 prezentac\u00ed ur\u010dit\u00e9ho postupu \u010di&nbsp;postup\u016f, kter\u00e9 jsou vhodn\u00e9 pro dostupn\u00e1 data, a pom\u00edj\u00ed tak postupy, kter\u00e9 nelze na vyu\u017e\u00edvan\u00fdch datech uspokojuj\u00edc\u00edm zp\u016fsobem vyu\u017e\u00edt.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">D\u00edky povaze dat OKpr\u00e1ce se diskuze metod a postup\u016f dopadov\u00e9 evaluace odehr\u00e1v\u00e1 v&nbsp;r\u00e1mci kvantitativn\u00edho kontrafaktu\u00e1ln\u00edho paradigmatu, je\u017e je zalo\u017eeno na jednofaktorov\u00e9m pojet\u00ed kauzality (kdy je na\u0161\u00ed snahou prost\u0159ednictv\u00edm kontrafaktu\u00e1lu izolovat efekt programu na odchod z nezam\u011bstnanosti od efektu jin\u00fdch faktor\u016f) a z\u00e1rove\u0148 dopad m\u011b\u0159\u00edme kvantitativn\u011b prost\u0159ednictv\u00edm statistick\u00e9 anal\u00fdzy. Z&nbsp;jin\u00fdch p\u0159\u00edstup\u016f zde zm\u00edn\u00edme pouze konfigura\u010dn\u00ed p\u0159\u00edstup, nap\u0159. Booleho anal\u00fdzu (Suchanec 2011), zalo\u017een\u00fd na komplexn\u00edm (v\u00edcefaktorov\u00e9m a interak\u010dn\u00edm) pojet\u00ed kauzality<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn4\"><sup>[4]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V&nbsp;textu se d\u00e1le zab\u00fdv\u00e1me n\u00e1sleduj\u00edc\u00edmi z\u00e1kladn\u00edmi oblastmi evalua\u010dn\u00edho postupu: a) definic\u00ed hodnocen\u00fdch osob a program\u016f, b) postupem stanoven\u00ed rozsahu a c\u00edlenosti programu, c) definic\u00ed v\u00fdsledku programu, d) hodnocen\u00edm dopadu, e) probl\u00e9mem selekce a jeho d\u016fsledky, f) p\u00e1rov\u00e1n\u00edm p\u0159\u00edpad\u016f (matching), g) probl\u00e9mem p\u0159eb\u011bhnut\u00ed a \u0159e\u0161en\u00edm nedokon\u010den\u00e9 \u00fa\u010dasti v&nbsp;programu, h) definic\u00ed \u010dasov\u00fdch bod\u016f pro m\u011b\u0159en\u00ed, i) diskuz\u00ed p\u016fsoben\u00ed \u010dasu a jeho vlivu na evaluaci program\u016f a j) technikami vyhodnocen\u00ed v\u00fdsledku. Uveden\u00e9 oblasti p\u0159edstavuj\u00ed podle na\u0161eho n\u00e1zoru kl\u00ed\u010dov\u00e1 metodologick\u00e1 m\u00edsta d\u016fle\u017eit\u00e1 pro vhodnou volbu konkr\u00e9tn\u00edho evalua\u010dn\u00edho postupu. Aplika\u010dn\u00ed potenci\u00e1l textu spat\u0159ujeme p\u0159edev\u0161\u00edm v jeho vyu\u017eit\u00ed p\u0159i&nbsp;tvorb\u011b konkr\u00e9tn\u00edho evalua\u010dn\u00edho postupu, konkr\u00e9tn\u011b p\u0159i p\u0159\u00edprav\u011b dat pro hodnocen\u00ed, volb\u011b vhodn\u00e9ho zp\u016fsobu p\u00e1rov\u00e1n\u00ed p\u0159\u00edpad\u016f a volb\u011b vhodn\u00e9 techniky pro vlastn\u00ed anal\u00fdzu.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Definice hodnocen\u00fdch osob a program\u016f<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Definice programu v&nbsp;prvn\u00ed \u0159ad\u011b zahrnuje vymezen\u00ed jednotliv\u00fdch program\u016f pro hodnocen\u00ed a jejich pro evaluaci v\u00fdznamn\u00fdch aspekt\u016f. Vymezen\u00ed program\u016f podle jednotliv\u00fdch typ\u016f a sub-typ\u016f bylo ji\u017e v&nbsp;odborn\u00e9 literatu\u0159e mnohokr\u00e1t prezentov\u00e1no jak v&nbsp;\u010cR, tak v&nbsp;zahrani\u010d\u00ed (viz nap\u0159. Bonoli 2010, Hora a Suchanec 2014) a nebudeme je zde opakovat. Upozorn\u00edme jen na&nbsp;n\u011bkter\u00e9 v\u00fdznamn\u00e9 specifick\u00e9 aspekty hodnocen\u00ed.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>V&nbsp;evaluaci na datech OKpr\u00e1ce se zam\u011b\u0159ujeme na osoby, kter\u00e9 byly ve sledovan\u00e9m obdob\u00ed v&nbsp;evidenci \u00da\u0159adu pr\u00e1ce.<\/li>\n\n\n\n<li>Proto\u017ee m\u00e1me k&nbsp;dispozici \u00fadaje o v\u0161ech nezam\u011bstnan\u00fdch, v\u0161ech programech a sou\u010dasn\u011b i informace o regionech a poskytovatel\u00edch program\u016f, je mo\u017en\u00e9 identifikovat region\u00e1ln\u00ed \u010di jin\u00e9 rozd\u00edly<br>mezi programy a vybrat vhodn\u00e9 ne\u00fa\u010dastn\u00edky pro kontroln\u00ed skupinu. Tato informace m\u016f\u017ee b\u00fdt v\u00fdznamn\u00e1, pokud se dopady program\u016f li\u0161\u00ed region\u00e1ln\u011b \u010di u jednotliv\u00fdch poskytovatel\u016f.<\/li>\n\n\n\n<li>Data n\u00e1m umo\u017e\u0148uj\u00ed hodnotit v\u0161echny programy bez nutnosti definice sub-vzork\u016f pro anal\u00fdzu, pokud je nechceme definovat vzhledem ke specifick\u00fdm c\u00edl\u016fm anal\u00fdzy<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn5\"><sup>[5]<\/sup><\/a>.<\/li>\n\n\n\n<li>Jednotliv\u00e9 programy mus\u00ed m\u00edt definov\u00e1n skute\u010dn\u00fd po\u010d\u00e1tek a konec intervence.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Specifick\u00fdm probl\u00e9mem pro evaluaci je opakovan\u00e1 (v\u00edcen\u00e1sobn\u00e1) \u00fa\u010dast nezam\u011bstnan\u00fdch v&nbsp;programech APZ ve sledovan\u00e9m obdob\u00ed, kter\u00e1 byla pops\u00e1na v&nbsp;\u010cR i v zahrani\u010d\u00ed (Hujer a Caliendo 2000, Richardson a van den Berg 2006, Hora a Suchanec 2014). V&nbsp;n\u011bkter\u00fdch p\u0159\u00edpadech je obdobn\u00fdm probl\u00e9mem zapojen\u00ed hodnocen\u00e9ho programu do \u0161ir\u0161\u00ed skupiny intervenc\u00ed \u2013 bal\u00ed\u010dku aktivit\/slu\u017eeb (Fay 1996).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Metodologick\u00fd postup stanoven\u00ed rozsahu a c\u00edlenosti programu<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Z&nbsp;hlediska dopadov\u00e9 evaluace je v\u00fdznamn\u00e9 p\u0159edev\u0161\u00edm stanoven\u00ed c\u00edlenosti (tedy v\u00fdsledn\u00e9ho zam\u011b\u0159en\u00ed program\u016f na jednotliv\u00e9 nezam\u011bstnan\u00e9 v z\u00e1vislosti na jejich charakteristik\u00e1ch). I kdy\u017e je probl\u00e9m c\u00edlenosti implicitn\u011b \u0159e\u0161en p\u00e1rov\u00e1n\u00edm p\u0159\u00edpad\u016f (viz n\u00ed\u017ee), m\u00e1 explicitn\u00ed vyj\u00e1d\u0159en\u00ed c\u00edlenosti dal\u0161\u00ed p\u0159\u00ednosy. Za prv\u00e9 m\u016f\u017ee b\u00fdt pro politick\u00e9 \u010dinitele vod\u00edtkem ohledn\u011b zam\u011b\u0159en\u00ed jednotliv\u00fdch program\u016f, za druh\u00e9 pak vytv\u00e1\u0159\u00ed z\u00e1kladn\u00ed vod\u00edtko pro reflektovanou interpretaci dopadu programu (k problematice c\u00edlenosti viz nap\u0159. Soukup 2006, Soukup, Michali\u010dka a Kot\u00edkov\u00e1 2009, Hora et al. 2009).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Definice v\u00fdsledku programu<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V&nbsp;t\u00e9to evaluaci je z\u00e1kladn\u00edm hodnot\u00edc\u00edm krit\u00e9riem posouzen\u00ed dopadu jednotliv\u00fdch program\u016f. V\u00fdchodiskem hodnocen\u00ed je shoda na definici o\u010dek\u00e1van\u00e9ho v\u00fdsledku program\u016f APZ. Mohr (1992) definuje v\u00fdsledek n\u00e1sleduj\u00edc\u00edm zp\u016fsobem:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>reaguje na n\u011bjak\u00fd spole\u010densk\u00fd probl\u00e9m,<\/li>\n\n\n\n<li>je n\u011bjak spole\u010densky cen\u011bn\u00fd,<\/li>\n\n\n\n<li>jedn\u00e1 se zpravidla o stav a nikoliv o aktivitu.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Podle Wimera (2006) je mo\u017en\u00e9 v\u00fdsledky d\u011blit na pozitivn\u00ed, neutr\u00e1ln\u00ed a&nbsp;negativn\u00ed, a to jak z&nbsp;matematick\u00e9ho, tak z&nbsp;normativn\u00edho hlediska. Tento p\u0159\u00edstup je hojn\u011b vyu\u017e\u00edv\u00e1n v&nbsp;meta-analytick\u00fdch studi\u00edch (nap\u0159. Calmfors, Forslund a Hemstr\u00f6m 2002, de Koning a Peers 2007, Card, Kluve a Weber 2009). V\u00fdsledek programu poskytuje prvn\u00ed orienta\u010dn\u00ed informaci o \u00fasp\u011b\u0161nosti programu a je z\u00e1kladn\u00edm p\u0159edpokladem pro hodnocen\u00ed dopadu programu<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn6\"><sup>[6]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V\u00fdsledky m\u016f\u017eeme rozli\u0161it tak\u00e9 na individu\u00e1ln\u00ed, skupinov\u00e9 a spole\u010densk\u00e9. Vliv programu nen\u00ed v&nbsp;jednotliv\u00fdch p\u0159\u00edpadech mo\u017en\u00e9 uspokojiv\u011b p\u0159edpov\u011bd\u011bt (Kluve et al. 2005), nebo\u0165 je kontra-faktu\u00e1ln\u00ed. Z\u00e1rove\u0148 je mo\u017en\u00e9 p\u0159edpokl\u00e1dat, \u017ee efekt programu se u jednotliv\u00fdch \u00fa\u010dastn\u00edk\u016f odli\u0161uje, a je proto z\u00e1visl\u00fd na za\u0159azen\u00ed konkr\u00e9tn\u00edch osob (Bryson, Dorsett a Purdon 2002). Hlavn\u00edm p\u0159edm\u011btem z\u00e1jmu v&nbsp;t\u00e9to dopadov\u00e9 perspektiv\u011b je proto&nbsp;pr\u016fm\u011brn\u00fd efekt jednotliv\u00fdch program\u016f na \u00fa\u010dastn\u00edky t\u011bchto program\u016f (viz&nbsp;nap\u0159. Borland, Tseng a Wilkins 2005, Dias, Ichimura a van den Berg 2008, Suchanec 2014a)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn7\"><sup>[7]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V evalua\u010dn\u00edch studi\u00edch (Betcherman, Olivas a Dar 2004, Kluve et al. 2005, Hujer, Thomsen a Zeiss 2006, de Koning a Peers 2007) m\u016f\u017eeme nal\u00e9zt nap\u0159. n\u00e1sleduj\u00edc\u00ed druhy v\u00fdsledk\u016f APZ:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>odchod z&nbsp;evidence\/nep\u0159\u00edtomnost v evidenci \u00fa\u0159adu pr\u00e1ce,<\/li>\n\n\n\n<li>zkr\u00e1cen\u00ed nezam\u011bstnanosti,<\/li>\n\n\n\n<li>nalezen\u00ed zam\u011bstn\u00e1n\u00ed,<\/li>\n\n\n\n<li>udr\u017een\u00ed (stabilita) zam\u011bstn\u00e1n\u00ed,<\/li>\n\n\n\n<li>zv\u00fd\u0161en\u00ed po\u010dtu odpracovan\u00fdch hodin,<\/li>\n\n\n\n<li>zv\u00fd\u0161en\u00ed produktivity,<\/li>\n\n\n\n<li>zv\u00fd\u0161en\u00ed mzdy (p\u0159\u00edjm\u016f),<\/li>\n\n\n\n<li>pracovn\u00ed podm\u00ednky vytvo\u0159en\u00fdch zam\u011bstn\u00e1n\u00ed,<\/li>\n\n\n\n<li>soci\u00e1ln\u00ed participace,<\/li>\n\n\n\n<li>zdravotn\u00ed stav.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">N\u011bkter\u00e9 evalua\u010dn\u00ed studie se zam\u011b\u0159ily na d\u00edl\u010d\u00ed v\u00fdsledky, kter\u00e9 jsou mezikroky k&nbsp;dosa\u017een\u00ed kone\u010dn\u00fdch v\u00fdsledk\u016f, nap\u0159. na hleda\u010dskou aktivitu nebo na reakce zam\u011bstnavatel\u016f na \u00fa\u010dast nezam\u011bstnan\u00fdch v&nbsp;programu (Calmfors, Forslund a Hemstr\u00f6m 2002). V\u00fdsledek je zpravidla d\u00e1le definov\u00e1n v&nbsp;r\u00e1mci vz\u00e1jemn\u011b se vylu\u010duj\u00edc\u00edch kategori\u00ed (nap\u0159. zam\u011bstnan\u00fd, nezam\u011bstnan\u00fd, ekonomicky neaktivn\u00ed). Kvalita jeho m\u011b\u0159en\u00ed pak z\u00e1vis\u00ed na&nbsp;schopnosti \u201espr\u00e1vn\u00e9ho\u201c p\u0159i\u0159azen\u00ed jednotliv\u00fdch p\u0159\u00edpad\u016f do t\u011bchto kategori\u00ed (viz Kluve et al. 2005). V&nbsp;dynamick\u00fdch modelech nemus\u00ed b\u00fdt v\u00fdsledek vyj\u00e1d\u0159en jako stav. Vermunt (1996) upozor\u0148uje, \u017ee tyto modely sp\u00ed\u0161e pracuj\u00ed s pojmy zm\u011bna (change), p\u0159echod (transition) nebo ud\u00e1lost (event). V\u00fdsledek je pak nap\u0159. m\u00edra v\u00fdskytu ud\u00e1losti (viz n\u00ed\u017ee zp\u016fsob hodnocen\u00ed v\u00fdsledku).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">D\u016fle\u017eit\u00fdm aspektem evaluac\u00ed zam\u011b\u0159en\u00fdch na odchod z&nbsp;evidence je mo\u017enost rozpoznat typ odchodu: do zam\u011bstn\u00e1n\u00ed, do podnik\u00e1n\u00ed, do vzd\u011bl\u00e1v\u00e1n\u00ed, do ekonomick\u00e9 neaktivity, sank\u010dn\u00edm vy\u0159azen\u00edm \u010di dokonce ztr\u00e1tou kontaktu mezi nezam\u011bstnan\u00fdm a pracovn\u00edkem \u00fa\u0159adu pr\u00e1ce (viz nap\u0159. Sianesi 2003). Podle Card, Kluve a Weber (2009) se zde pracuje s (implicitn\u00edm a normativn\u00edm) p\u0159edpokladem, \u017ee odchod do zam\u011bstnanosti je \u201edobr\u00fdm\u201c typem v\u00fdsledku, zat\u00edmco odchod z&nbsp;jin\u00fdch d\u016fvod\u016f je \u201e\u0161patn\u00fd\u201c v\u00fdsledek. Do t\u00e9to \u00favahy vstupuje nap\u0159. n\u00e1sleduj\u00edc\u00ed:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>c\u00edle politik,<\/li>\n\n\n\n<li>ti, kdo ode\u0161li z&nbsp;registru \u00da\u0159adu pr\u00e1ce, nejsou p\u0159\u00edjemci soci\u00e1ln\u00edch d\u00e1vek (Sianesi 2003),<\/li>\n\n\n\n<li>\u0161ir\u0161\u00ed hlediska ekonomick\u00e9ho vlivu na ve\u0159ejn\u00e9 rozpo\u010dty,<\/li>\n\n\n\n<li>d\u016fraz na etiku pr\u00e1ce.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Card, Kluve a Weber (2009) uv\u00e1d\u011bj\u00ed, \u017ee evalua\u010dn\u00ed studie, kter\u00e9 jsou zam\u011b\u0159eny na v\u00fdsledek \u201ed\u00e9lku registrovan\u00e9 nezam\u011bstnanosti\u201c, p\u0159in\u00e1\u0161ej\u00ed pozitivn\u011bj\u0161\u00ed kr\u00e1tkodob\u00e9 v\u00fdsledky ne\u017e studie, kter\u00e9 se zam\u011b\u0159uj\u00ed na v\u00fdsledek \u201ezam\u011bstnanost\u201c nebo \u201ep\u0159\u00edjem\u201c (pro konkr\u00e9tn\u00ed p\u0159\u00edklad rozd\u00edlnosti v\u00fdsledku viz Dias, Ichimura a van den Berg 2008).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">D\u016fsledkem chyb\u011bj\u00edc\u00edch informac\u00ed o v\u00fdsledku n\u011bkter\u00fdch nezam\u011bstnan\u00fdch je riziko \u0161patn\u00e9ho p\u0159i\u0159azen\u00ed a t\u00edm i odhadu dopadu programu (Sianesi 2003). Sou\u010dasn\u011b hroz\u00ed riziko, \u017ee pravd\u011bpodobnost chyb\u011bj\u00edc\u00ed informace se systematicky odli\u0161uje mezi skupinou intervence a kontroln\u00ed skupinou (Sianesi 2003). V&nbsp;n\u011bkter\u00fdch evalua\u010dn\u00edch studi\u00edch auto\u0159i pracuj\u00ed s chyb\u011bj\u00edc\u00edmi v\u00fdsledky jako s&nbsp;cenzorovan\u00fdmi p\u0159\u00edpady, jindy se pokou\u0161ej\u00ed o odhad m\u00edry odchodu do zam\u011bstn\u00e1n\u00ed. Bring a Carling (2001) \u0159e\u0161ili ot\u00e1zku nezn\u00e1m\u00e9ho d\u016fvodu odchodu nezam\u011bstnan\u00fdch z&nbsp;evidence proveden\u00edm dodate\u010dn\u00e9ho telefonn\u00edho \u0161et\u0159en\u00ed a zjistili, \u017ee v\u00edce ne\u017e polovina t\u011bchto osob ode\u0161la do ekonomick\u00e9 neaktivity. D\u00e1le nepochybn\u011b z\u00e1le\u017e\u00ed na tom, jak velk\u00e1 proporce ekonomicky neaktivn\u00edch se pozd\u011bji vr\u00e1t\u00ed do zam\u011bstn\u00e1n\u00ed (v p\u0159\u00edpad\u011b Bring a Carling 2001 byla tato proporce velmi mal\u00e1)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn8\"><sup>[8]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Je t\u0159eba br\u00e1t v&nbsp;\u00favahu, \u017ee \u00fa\u010dast v&nbsp;programu m\u016f\u017ee m\u00edt o\u010dek\u00e1van\u00e9 i neo\u010dek\u00e1van\u00e9 vedlej\u0161\u00ed v\u00fdsledky. Calmfors, Forslund a Hemstr\u00f6m (2002) a Hujer, Thomsen a Zeiss (2006) upozor\u0148uj\u00ed na to, \u017ee programy mohou sou\u010dasn\u011b vytv\u00e1\u0159et protich\u016fdn\u00e9 efekty (nap\u0159. podporovat i potla\u010dovat motivaci k&nbsp;hled\u00e1n\u00ed zam\u011bstn\u00e1n\u00ed). \u010cast\u00fdm takov\u00fdm d\u016fsledkem u n\u011bkter\u00fdch typ\u016f program\u016f je efekt uzam\u010den\u00ed (locking-in) nebo automatick\u00e9 participace. \u00da\u010dastn\u00edci vzd\u011bl\u00e1vac\u00edch program\u016f mohou m\u00edt b\u011bhem trv\u00e1n\u00ed programu men\u0161\u00ed aktivitu p\u0159i hled\u00e1n\u00ed zam\u011bstn\u00e1n\u00ed a men\u0161\u00ed re\u00e1lnou m\u00edru zam\u011bstnanosti ne\u017e ne\u00fa\u010dastn\u00edci, naopak \u00fa\u010dastn\u00edci program\u016f tvorby m\u00edst jsou \u010dasto automaticky ch\u00e1p\u00e1ni jako zam\u011bstnan\u00ed \u010di nezam\u011bstnan\u00ed (de Koning a Peers 2007, Card, Ibarr\u00e1n a Villa 2011, Forslund, Fredriksson a Vinkstr\u00f6m 2011). Z&nbsp;tohoto d\u016fvodu je nutn\u00e9 dos\u00e1hnout shody na definici v\u00fdsledku a odli\u0161en\u00ed jednotliv\u00fdch stav\u016f, a to p\u0159edev\u0161\u00edm u program\u016f tvorby m\u00edst.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Richardson a van den Berg (2006) na konkr\u00e9tn\u00edm programu testovali vliv efektu uzam\u010den\u00ed. Zjistili, \u017ee i kdy\u017e byl p\u0159i nezapo\u010dten\u00ed d\u00e9lky programu efekt dlouhodob\u011bj\u0161\u00edch program\u016f v\u011bt\u0161\u00ed ne\u017e efekt program\u016f kr\u00e1tkodob\u00fdch, tento efekt byl zcela potla\u010den v&nbsp;p\u0159\u00edpad\u011b, \u017ee se zapo\u010detla d\u00e9lka programu (viz ibid.). Hora a Suchanec (2014) uk\u00e1zali, \u017ee u \u00fa\u010dastn\u00edk\u016f v\u00edce ne\u017e jednoho programu byly dopady program\u016f men\u0161\u00ed ne\u017e u \u00fa\u010dastn\u00edk\u016f jedin\u00e9ho programu. Posouzen\u00ed efektu programu v&nbsp;takov\u00e9m p\u0159\u00edpad\u011b vych\u00e1z\u00ed z&nbsp;normativn\u00edch hledisek<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn9\"><sup>[9]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Dal\u0161\u00edm p\u0159\u00edkladem efektu, kter\u00fd je mo\u017en\u00e9 podle n\u011bkter\u00fdch autor\u016f (Suchanec 2014a) identifikovat na z\u00e1klad\u011b dat v&nbsp;perspektiv\u011b \u010d\u00e1ste\u010dn\u00e9ho ekvilibria, je tzv. efekt mrtv\u00e9 v\u00e1hy. Tito auto\u0159i odhaduj\u00ed efekt mrtv\u00e9 v\u00e1hy<br>jako&nbsp;pod\u00edl osob v&nbsp;kontroln\u00ed skupin\u011b, kter\u00e9 si nalezly zam\u011bstn\u00e1n\u00ed i bez \u00fa\u010dasti v&nbsp;programu tvorby m\u00edst.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Je t\u0159eba poznamenat, \u017ee anal\u00fdza v&nbsp;perspektiv\u011b \u010d\u00e1ste\u010dn\u00e9ho ekvilibria m\u016f\u017ee v\u00e9st ke zkreslen\u00e9mu odhadu, pokud je neadekv\u00e1tn\u011b zobecn\u011bna. Tento p\u0159\u00edstup toti\u017e nen\u00ed schopen odhalit \u010d\u00e1st popsan\u00fdch vedlej\u0161\u00edch efekt\u016f APZ na cel\u00fd trh pr\u00e1ce (srovnej nap\u0159. Calmfors 1994).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V&nbsp;sumativn\u00ed anal\u00fdze nen\u00ed v\u00fdsledek (na rozd\u00edl od n\u011bkter\u00fdch jin\u00fdch na procesn\u00ed evaluaci zalo\u017een\u00fdch p\u0159\u00edstup\u016f) dostate\u010dn\u00fdm c\u00edlem evaluace, ale auto\u0159i usiluj\u00ed o zji\u0161t\u011bn\u00ed dopadu (viz n\u00ed\u017ee).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Hodnocen\u00ed dopadu \u2013 zp\u016fsob \u0159e\u0161en\u00ed kontra-faktu\u00e1ln\u00edho probl\u00e9mu<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Evaluace vych\u00e1z\u00ed z o\u010dek\u00e1v\u00e1n\u00ed kauz\u00e1ln\u00edho vztahu mezi \u00fa\u010dast\u00ed v&nbsp;programu APZ a definovan\u00fdm v\u00fdsledkem. Dopad programu je podle Borland, Tseng a Wilkins (2005) definov\u00e1n takto: nakolik se zm\u011bn\u00ed m\u011b\u0159en\u00fd v\u00fdsledek participac\u00ed v&nbsp;programu\u2026<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn10\"><sup>[10]<\/sup><\/a>. Z\u00e1kladn\u00edm evalua\u010dn\u00edm probl\u00e9mem je ov\u0161em odli\u0161en\u00ed vlivu programu od vlivu dal\u0161\u00edch vn\u011bj\u0161\u00edch faktor\u016f (Hujer, Caliendo a Radi\u0107 2004). Nejistota panuje p\u0159edev\u0161\u00edm ohledn\u011b toho:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>zda je v\u00fdsledek skute\u010dn\u011b zp\u016fsoben programem a ne n\u011b\u010d\u00edm jin\u00fdm, nap\u0159. kompetencemi jednotliv\u00fdch nezam\u011bstnan\u00fdch,<\/li>\n\n\n\n<li>zda by se nezam\u011bstnan\u00fdm nevedlo l\u00e9pe bez programu (zda by si pr\u00e1ci nenalezli, i pokud by se programu nez\u00fa\u010dastnili).<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Z&nbsp;tohoto d\u016fvodu by bylo \u017e\u00e1douc\u00ed porovnat situaci stejn\u00e9ho \u00fa\u010dastn\u00edka p\u0159i&nbsp;\u00fa\u010dasti v&nbsp;programu se situac\u00ed bez \u00fa\u010dasti v&nbsp;programu, co\u017e ov\u0161em nelze. Efekt \u010di dopad programu tedy p\u0159edstavuje pomysln\u00fd rozd\u00edl mezi dosa\u017een\u00fdm v\u00fdsledkem \u00fa\u010dastn\u00edk\u016f programu a jejich kontra-faktu\u00e1ln\u00edm (rozum\u011bj nenastal\u00fdm) v\u00fdsledkem, pokud by se programu nebyli \u00fa\u010dastnili (viz Hujer a Wellner 2000, de Koning a Peers 2007, Card, Ibarr\u00e1n a Villa 2011).<br>Proto\u017ee p\u0159\u00edm\u00e9 pozorov\u00e1n\u00ed obou stav\u016f sou\u010dasn\u011b nen\u00ed mo\u017en\u00e9, je efekt programu odhadov\u00e1n (Hujer a Wellner 2000, Betcherman, Olivas a Dar 2004, Card, Ibarr\u00e1n a Villa 2011). K&nbsp;odhadu (proxy) v\u00fdsledku skupiny intervence v&nbsp;situaci bez programu se nej\u010dast\u011bji vyu\u017e\u00edv\u00e1 v\u00fdsledku, kter\u00e9ho dos\u00e1hli ne\u00fa\u010dastn\u00edci programu (Hujer, Caliendo a Radi\u0107 2004). Tato situace vy\u017eaduje tzv. definov\u00e1n\u00ed identifika\u010dn\u00edch p\u0159edpoklad\u016f (Bryson, Dorsett a&nbsp;Purdon 2002), tedy podm\u00ednek, kter\u00e9 mus\u00ed b\u00fdt spln\u011bny, aby byly v\u00fdsledky evaluace platn\u00e9 (viz t\u00e9\u017e n\u00ed\u017ee).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Ve v\u011bt\u0161in\u011b evalua\u010dn\u00edch studi\u00ed doch\u00e1z\u00ed k&nbsp;odhadu parametru populace z&nbsp;men\u0161\u00edho v\u00fdb\u011brov\u00e9ho vzorku s&nbsp;rizikem nep\u0159esnosti odhadu (viz Kluve et&nbsp;al. 2005). Mohr (1992) definuje citlivost designu jako m\u011b\u0159\u00edtko toho, jak&nbsp;mal\u00fd m\u016f\u017ee b\u00fdt rozd\u00edl ve v\u00fdsledku mezi experiment\u00e1ln\u00ed skupinou a&nbsp;kontrafaktu\u00e1ln\u00ed skupinou, abychom st\u00e1le v\u011b\u0159ili, \u017ee program m\u00e1 pozitivn\u00ed kauz\u00e1ln\u00ed dopad. Rozd\u00edl, kter\u00fd jsme schopni detekovat, z\u00e1le\u017e\u00ed na p\u0159esnosti na\u0161eho odhadu a \u0159ad\u011b dal\u0161\u00edch faktor\u016f (mj. velikost vzorku). Citlivost designu je ov\u0161em v\u00fdznamn\u00e1 pouze tam, kde \u00fa\u010dastn\u00edky programu vyb\u00edr\u00e1me z&nbsp;populace nezam\u011bstnan\u00fdch n\u00e1hodn\u011b p\u0159i snaze zobecnit na tuto populaci, co\u017e v&nbsp;p\u0159\u00edpad\u011b administrativn\u00edch dat OKpr\u00e1ce neplat\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jeliko\u017e v p\u0159\u00edpad\u011b anal\u00fdzy na datech OKpr\u00e1ce nelze \u00fa\u010dastn\u00edky a ne\u00fa\u010dastn\u00edky rozd\u011blit do obou skupin n\u00e1hodn\u011b (jak je tomu v&nbsp;klasick\u00e9m experimentu), vych\u00e1z\u00ed konstrukce obou skupin pro m\u011b\u0159en\u00ed dopadu dodate\u010dn\u011b z&nbsp;rozhodnut\u00ed v\u00fdzkumn\u00edk\u016f na z\u00e1klad\u011b skute\u010dn\u00fdch dat. Tento postup je n\u011bkter\u00fdmi autory ozna\u010dov\u00e1n za kvazi-experiment\u00e1ln\u00ed design v\u00fdzkumu, zat\u00edmco jin\u00ed auto\u0159i jej ch\u00e1pou jako design studie na z\u00e1klad\u011b pozorov\u00e1n\u00ed. Hlavn\u00ed v\u00fdhodou takov\u00e9ho p\u0159\u00edstupu je n\u00edzk\u00e1 m\u00edra rizika naru\u0161en\u00ed prost\u0159ed\u00ed v\u00fdzkumem. Z\u00e1sadn\u00ed nev\u00fdhodou je ov\u0161em pot\u0159eba vypo\u0159\u00e1dat se s probl\u00e9mem selekce a dal\u0161\u00edmi probl\u00e9my, kter\u00e9 vypl\u00fdvaj\u00ed z&nbsp;povahy dostupn\u00fdch dat.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Probl\u00e9m selekce a jeho d\u016fsledky<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Dopad programu je validn\u011b odhadnut pouze, pokud nedoch\u00e1z\u00ed k&nbsp;tzv.&nbsp;zkreslen\u00ed na z\u00e1klad\u011b selekce (Hujer a Wellner 2000). Probl\u00e9m zkreslen\u00ed na z\u00e1klad\u011b selekce vznik\u00e1 (p\u0159i nep\u0159\u00edtomnosti n\u00e1hodn\u00e9ho v\u00fdb\u011bru \u00fa\u010dastn\u00edk\u016f a ne\u00fa\u010dastn\u00edk\u016f programu) v&nbsp;d\u016fsledku existence r\u016fzn\u00fdch okolnost\u00ed vedouc\u00edch k&nbsp;\u00fa\u010dasti \u010di ne\u00fa\u010dasti jednotliv\u00fdch nezam\u011bstnan\u00fdch v programech APZ. Takov\u00fdmi okolnostmi mohou b\u00fdt nap\u0159. podm\u00ednky n\u00e1roku pro vstup do programu, kdy jsou preferov\u00e1ni obt\u00ed\u017en\u011b um\u00edstiteln\u00ed uchaze\u010di, samo-v\u00fdb\u011br \u00fa\u010dastn\u00edk\u016f do programu, v\u00fdb\u011br na z\u00e1klad\u011b tzv. efektu sl\u00edz\u00e1v\u00e1n\u00ed smetany (cream-skimming)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn11\"><sup>[11]<\/sup><\/a>&nbsp;\u010di p\u0159edpoklad n\u00e1stupu do zam\u011bstn\u00e1n\u00ed (Bryson, Dorsett a Purdon 2002, Lechner a Wunsch 2009, Card, Ibarr\u00e1n a&nbsp;Villa 2011).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jako z\u00e1kladn\u00ed krit\u00e9ria posouzen\u00ed rizika probl\u00e9mu selekce jsou br\u00e1ny nap\u0159.&nbsp;(ne)jednotnost a (ne)centralita za\u0159azov\u00e1n\u00ed \u010di (ne)za\u0159azov\u00e1n\u00ed podle jasn\u00fdch krit\u00e9ri\u00ed<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn12\"><sup>[12]<\/sup><\/a>. P\u0159i nedodr\u017een\u00ed t\u011bchto hledisek roste nejistota ohledn\u011b skryt\u00fdch faktor\u016f p\u016fsob\u00edc\u00edch na r\u016fznost obou skupin a t\u00edm i pochybnost o&nbsp;shodnosti obou skupin. V&nbsp;d\u016fsledku selekce je rozd\u00edl mezi v\u00fdsledky \u00fa\u010dastn\u00edk\u016f a ne\u00fa\u010dastn\u00edk\u016f programu zp\u016fsoben jak samotn\u00fdm programem, tak sou\u010dasn\u011b \u010di v\u00fdhradn\u011b \u0159adou dal\u0161\u00edch potencion\u00e1ln\u00edch faktor\u016f (viz Mohr 1992, Card, Ibarr\u00e1n a Villa 2011). Praktick\u00fd p\u0159\u00edklad d\u016fsledku neprav\u00e9ho efektu v d\u016fsledku selekce uv\u00e1d\u00ed Richardson a van den Berg (2006). Pokud \u00fa\u010dastn\u00edk po programu pom\u011brn\u011b rychle opust\u00ed evidenci, m\u016f\u017ee to b\u00fdt v&nbsp;d\u016fsledku a) pozitivn\u00edho kauz\u00e1ln\u00edho efektu programu nebo b) skryt\u00fdch faktor\u016f, kter\u00e9 by vedly k&nbsp;rychl\u00e9mu odchodu z&nbsp;evidence i bez programu. Skute\u010dn\u00fd rozsah probl\u00e9mu selekce p\u0159itom \u010dasto nen\u00ed v\u00fdzkumn\u00edkovi zn\u00e1m (Fay 1996).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V\u00fdzkumn\u00edk tedy nem\u016f\u017ee z&nbsp;v\u00fd\u0161e uveden\u00fdch d\u016fvod\u016f realisticky p\u0159edpokl\u00e1dat, \u017ee by k&nbsp;probl\u00e9mu selekce nedoch\u00e1zelo (mj. proto, \u017ee je dlouhodob\u011b zdokumentov\u00e1na c\u00edlenost program\u016f APZ na ur\u010dit\u00e9 skupiny, a tud\u00ed\u017e \u0161ance na vstup do programu rozhodn\u011b nen\u00ed pro v\u0161echny stejn\u00e1). V&nbsp;t\u00e9to metodice z\u00e1rove\u0148 vych\u00e1z\u00edme z&nbsp;p\u0159edpokladu, \u017ee v\u00fdzkumn\u00edk nem\u016f\u017ee v ne-experiment\u00e1ln\u00edm v\u00fdzkumu proces selekce efektivn\u011b ovlivnit<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn13\"><sup>[13]<\/sup><\/a>, a proto mus\u00ed o\u010dek\u00e1vat existenci p\u0159ed-programov\u00fdch rozd\u00edl\u016f mezi skupinou intervence a kontroln\u00ed skupinou a volit adekv\u00e1tn\u00ed \u0159e\u0161en\u00ed za pomoci korekce volbou vhodn\u00e9ho designu a estim\u00e1toru (Hujer a Caliendo 2000).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Kluve et al. (2005) d\u011bl\u00ed faktory s&nbsp;mo\u017en\u00fdm vlivem na v\u00fdsledky programu na: (a) sou\u010dasn\u00e9 charakteristiky a (b) p\u0159edchoz\u00ed v\u00fdsledky na trhu pr\u00e1ce. Kluve et al. (2005) uv\u00e1d\u00ed, \u017ee je d\u016fle\u017eit\u00e9, aby byly tyto faktory vn\u011bj\u0161\u00edmi faktory, tedy jde o to nepodmi\u0148ovat na faktorech, jejich\u017e zm\u011bna m\u016f\u017ee b\u00fdt v\u00fdsledkem participace v&nbsp;programu. Faktory, kter\u00e9 mohou zp\u016fsobovat odli\u0161nost skupin lze d\u00e1le rozd\u011blit na snadno zjistiteln\u00e9 (takov\u00e9, kter\u00e9 lze ve&nbsp;form\u011b prom\u011bnn\u00fdch zahrnout do v\u00fdzkumn\u00e9ho souboru a lze pro n\u011b kontrolovat) a obt\u00ed\u017en\u011b zjistiteln\u00e9 \u010di nezjistiteln\u00e9. Toto rozd\u011blen\u00ed m\u00e1 z\u00e1sadn\u00ed v\u00fdznam pro identifikaci vhodn\u00e9ho zp\u016fsobu \u0159e\u0161en\u00ed probl\u00e9mu selekce. Probl\u00e9m nepozorovan\u00e9 heterogenity v&nbsp;d\u016fsledku neza\u0159azen\u00ed prom\u011bnn\u00fdch do&nbsp;modelu jsou z\u00e1va\u017en\u00fdm rizikem i u model\u016f hazardu (Vermunt 1996). Bryson, Dorsett a Purdon (2002) poukazuj\u00ed na to, \u017ee zahrnut\u00edm pozorovan\u00fdch prom\u011bnn\u00fdch m\u016f\u017eeme \u010d\u00e1ste\u010dn\u011b zahrnout i vliv nepozorovan\u00fdch prom\u011bnn\u00fdch, pokud jsou tyto vz\u00e1jemn\u011b korelov\u00e1ny (nap\u0159. p\u0159ed-programov\u00e1 historie a motivace).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>\u0158e\u0161en\u00ed probl\u00e9mu selekce za pomoci p\u00e1rov\u00e1n\u00ed p\u0159\u00edpad\u016f (matching)<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">P\u00e1rov\u00e1n\u00ed p\u0159\u00edpad\u016f (\u010dasto v&nbsp;zahrani\u010d\u00ed ozna\u010dov\u00e1no jako tzv. matching), je jedn\u00edm z&nbsp;\u010dasto pou\u017e\u00edvan\u00fdch postup\u016f rozd\u011blen\u00ed vzorku na dv\u011b srovnateln\u00e9 skupiny: \u00fa\u010dastn\u00edky programu a ne\u00fa\u010dastn\u00edky programu<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn14\"><sup>[14]<\/sup><\/a>. Podstatou procesu p\u00e1rov\u00e1n\u00ed je v\u00fdb\u011br osob do kontroln\u00ed skupiny (z \u0161ir\u0161\u00edho okruhu osob) tak, aby p\u0159ed-programov\u00e9 charakteristiky t\u011bchto osob co nejv\u00edce odpov\u00eddaly charakteristik\u00e1m osob, kter\u00e9 se programu \u00fa\u010dastnily (de Koning a&nbsp;Peers 2007, Card, Ibarr\u00e1n a Villa 2011). Jin\u00fdmi slovy, c\u00edlem p\u00e1rov\u00e1n\u00ed je vybalancovat rozlo\u017een\u00ed v\u0161ech relevantn\u00edch p\u0159ed-programov\u00fdch charakteristik v&nbsp;obou skupin\u00e1ch, a t\u00edm dos\u00e1hnout nez\u00e1vislosti mezi potencion\u00e1ln\u00edmi v\u00fdsledky a za\u0159azen\u00edm do programu (Hujer, Caliendo a Radi\u0107 2004). Identifika\u010dn\u00ed p\u0159edpoklad podm\u00edn\u011bn\u00e9 nez\u00e1vislosti (conditional independence assumption \u2013 CIA) toti\u017e vy\u017eaduje, aby individu\u00e1ln\u00ed \u00fa\u010dast v programu byla nez\u00e1visl\u00e1 na potenci\u00e1ln\u00edm v\u00fdsledku programu v&nbsp;situaci bez&nbsp;programu (Sianesi 2003, Borland, Tseng a Wilkins 2005, Dias, Ichimura a van den Berg 2008). CIA je spln\u011bna pouze pokud v\u0161echny prom\u011bnn\u00e9, kter\u00e9 ovliv\u0148uj\u00ed jak proces selekce, tak potenci\u00e1ln\u00ed v\u00fdsledek programu, jsou zahrnuty do p\u00e1rov\u00e1n\u00ed (Bryson, Dorsett a Purdon 2002, Reinowski a&nbsp;Schultz 2006). Neidentifikovan\u00e9 faktory jsou t\u00e9\u017e potenci\u00e1ln\u011b nebezpe\u010dn\u00e9 pro zkreslen\u00ed v\u00fdsledk\u016f pouze tehdy, pokud sou\u010dasn\u011b souvisej\u00ed jak s participac\u00ed v&nbsp;programu, tak s&nbsp;potenci\u00e1ln\u00edm v\u00fdsledkem. Z&nbsp;v\u00fd\u0161e uveden\u00e9ho vypl\u00fdv\u00e1, \u017ee ke zkvalitn\u011bn\u00ed p\u00e1rov\u00e1n\u00ed doch\u00e1z\u00ed zahrnut\u00edm co nej\u0161ir\u0161\u00ed skupiny z&nbsp;v\u00fd\u0161e uveden\u00e9ho hlediska relevantn\u00edch prom\u011bnn\u00fdch.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">P\u00e1rov\u00e1n\u00ed je ov\u0161em zalo\u017eeno t\u00e9\u017e na p\u0159edpokladu spole\u010dn\u00e9 podpory (common support assumption \u2013 CSA), tedy \u017ee ke ka\u017ed\u00e9mu \u00fa\u010dastn\u00edkovi z\u00edsk\u00e1me shodn\u00e9ho \u010di obdobn\u00e9ho ne\u00fa\u010dastn\u00edka (Sianesi 2003, Borland, Tseng a Wilkins 2005, Reinowski a Schultz 2006). Rizikem procesu p\u00e1rov\u00e1n\u00ed je proto nar\u016fstaj\u00edc\u00ed po\u010det prom\u011bnn\u00fdch zahrnut\u00fdch do procesu p\u00e1rov\u00e1n\u00ed, nebo\u0165 s&nbsp;ka\u017edou dal\u0161\u00ed prom\u011bnnou exponenci\u00e1ln\u011b nar\u016fst\u00e1 i po\u010det bun\u011bk a t\u00edm mo\u017enost\u00ed, kde se p\u00e1rovan\u00e9 osoby mohou odli\u0161ovat (viz nap\u0159. Dehejia a Wahba 2002, Hujer, Caliendo a Radi\u0107 2004). CSA m\u016f\u017ee b\u00fdt probl\u00e9mem tak\u00e9 v&nbsp;p\u0159\u00edpad\u011b povinn\u00fdch program\u016f, nebo\u0165 nen\u00ed mo\u017en\u00e9 naj\u00edt odpov\u00eddaj\u00edc\u00ed ne\u00fa\u010dastn\u00edky (Bryson, Dorsett a Purdon 2002).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">P\u0159i p\u00e1rov\u00e1n\u00ed p\u0159\u00edpad\u016f se jako prom\u011bnn\u00e1 pro p\u00e1rov\u00e1n\u00ed pou\u017e\u00edv\u00e1 tzv. propensity score tedy podm\u00edn\u011bn\u00e1 pravd\u011bpodobnost pro \u00fa\u010dast v&nbsp;programu (Hujer a Wellner 2000, Dehejia a Wahba 2002). Velkou v\u00fdhodou tohoto postupu je mo\u017enost p\u00e1rovat na jedin\u00e9 prom\u011bnn\u00e9 (Hujer, Caliendo a Radi\u0107 2004). Dal\u0161\u00ed zna\u010dnou v\u00fdhodou je, \u017ee d\u00edky propensity score m\u016f\u017eeme p\u00e1rovat na velk\u00e9m mno\u017estv\u00ed prom\u011bnn\u00fdch bez rizika zkreslen\u00ed v\u00fdsledku (Dehejia a Wahba 2002). P\u00e1rov\u00e1n\u00ed je spojeno s&nbsp;rizikem zkreslen\u00ed p\u0159edev\u0161\u00edm v&nbsp;t\u011bch p\u0159\u00edpadech, kdy v&nbsp;kontroln\u00ed skupin\u011b nenal\u00e9z\u00e1me vhodn\u00e9 p\u0159\u00edpady pro sp\u00e1rov\u00e1n\u00ed (Hujer, Caliendo a Radi\u0107 2004). Z&nbsp;toho vypl\u00fdv\u00e1, \u017ee je v\u00fdhodou m\u00edt dostate\u010dn\u00fd po\u010det osob, ze kter\u00fdch je mo\u017en\u00e9 do kontroln\u00ed skupiny vyb\u00edrat. P\u0159\u00edpady, pro kter\u00e9 nem\u00e1me odpov\u00eddaj\u00edc\u00ed prot\u011bj\u0161ky v&nbsp;kontroln\u00ed skupin\u011b, nem\u016f\u017eeme zahrnout do procesu p\u00e1rov\u00e1n\u00ed. V&nbsp;n\u011bkter\u00fdch p\u0159\u00edpadech auto\u0159i kombinovali vyu\u017eit\u00ed propensity score s&nbsp;p\u0159esn\u00fdm p\u00e1rov\u00e1n\u00edm na konkr\u00e9tn\u00ed prom\u011bnn\u00e9 historie p\u0159ed programem (Puhani 1998, Hora a&nbsp;Suchanec 2014) \u2013 k&nbsp;d\u016fvod\u016fm pro vyu\u017eit\u00ed tohoto p\u0159\u00edstupu viz n\u00ed\u017ee definice \u010dasov\u00fdch bod\u016f.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Podle mo\u017enost\u00ed je ke ka\u017ed\u00e9mu \u00fa\u010dastn\u00edku programu vyhled\u00e1n jeden \u010di v\u00edce ne\u00fa\u010dastn\u00edk\u016f anebo je jeden ne\u00fa\u010dastn\u00edk vyu\u017eit i v\u00edcekr\u00e1t (Hujer, Caliendo a&nbsp;Radi\u0107 2004). Existuje v\u00edce zp\u016fsob\u016f vyu\u017eit\u00ed propensity score pro p\u00e1rov\u00e1n\u00ed na z\u00e1klad\u011b dostupnosti vhodn\u00fdch p\u0159\u00edpad\u016f pro p\u00e1rov\u00e1n\u00ed (viz pozn\u00e1mka)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn15\"><sup>[15]<\/sup><\/a>. P\u0159i vyu\u017eit\u00ed procesu p\u00e1rov\u00e1n\u00ed podle skupin jsou n\u011bkdy pou\u017eity v\u00e1hy, kter\u00e9 zaji\u0161\u0165uj\u00ed stejnou distribuci charakteristik ve skupin\u011b intervence a&nbsp;v&nbsp;kontroln\u00ed skupin\u011b podle po\u010dtu p\u0159\u00edpad\u016f (Kluve et al. 2005).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Vhodn\u00fdm zp\u016fsobem p\u00e1rov\u00e1n\u00ed na datech OKpr\u00e1ce m\u016f\u017ee b\u00fdt p\u00e1rov\u00e1n\u00ed bez&nbsp;n\u00e1hrady s n\u00e1hodn\u00fdm p\u0159i\u0159azen\u00edm osob z&nbsp;kontroln\u00ed skupiny (viz nap\u0159. Dehejia a Wahba 2002, Reinowski a Schultz 2006). P\u00e1rov\u00e1n\u00ed bez n\u00e1hrady v&nbsp;praxi znamen\u00e1, \u017ee ka\u017ed\u00fd p\u0159\u00edpad je za\u0159azen do p\u00e1rov\u00e1n\u00ed pouze jednou, tj.&nbsp;je-li vytvo\u0159en p\u00e1r, oba p\u0159\u00edpady jsou z&nbsp;dal\u0161\u00edho p\u00e1rov\u00e1n\u00ed vy\u0159azeny (Dehejia a Wahba 2002). To je pravd\u011bpodobn\u011b vhodn\u011bj\u0161\u00ed p\u0159\u00edstup, pokud m\u00e1me vysokou podobnost respektive mo\u017enost podobnosti kontroln\u00ed skupiny a&nbsp;skupiny intervence (viz Dehejia a Wahba 2002)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn16\"><sup>[16]<\/sup><\/a>. Nev\u00fdhodou tohoto p\u0159\u00edstupu je, \u017ee m\u016f\u017ee b\u00fdt citliv\u00fd na po\u0159ad\u00ed \u0159azen\u00fdch p\u0159\u00edpad\u016f (Dehejia a&nbsp;Wahba 2002). Velkou v\u00fdhodou dat OKpr\u00e1ce je vysok\u00fd po\u010det dostupn\u00fdch p\u0159\u00edpad\u016f pro p\u00e1rov\u00e1n\u00ed. V&nbsp;minul\u00fdch anal\u00fdz\u00e1ch bylo mo\u017en\u00e9 sp\u00e1rovat zpravidla 80\u201395 procent p\u0159\u00edpad\u016f. Zpravidla pro t\u00e9m\u011b\u0159 v\u0161echny p\u0159\u00edpady ve&nbsp;skupin\u011b intervence m\u00e1me vy\u0161\u0161\u00ed po\u010det potencion\u00e1ln\u00edch partner\u016f v kontroln\u00ed skupin\u011b. Tyto p\u0159\u00edpady lze p\u0159i\u0159adit ke skupin\u011b intervence pomoc\u00ed n\u00e1hodn\u00fdch \u010d\u00edsel. Alternativn\u00edm p\u0159\u00edstupem by mohlo b\u00fdt zam\u011b\u0159en\u00ed se na&nbsp;p\u00e1rov\u00e1n\u00ed dosud nesp\u00e1rovan\u00fdch p\u0159\u00edpad\u016f<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn17\"><sup>[17]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Nev\u00fdhodou procesu p\u00e1rov\u00e1n\u00ed (matching) je potenci\u00e1ln\u00ed neznalost skryt\u00fdch charakteristik (nap\u0159. motivace, aktivita p\u0159i hled\u00e1n\u00ed zam\u011bstn\u00e1n\u00ed, kompetence nezam\u011bstnan\u00fdch), kter\u00e9 mohou m\u00edt vliv jak na vstup do programu tak na \u0161ance nezam\u011bstnan\u00fdch na trhu pr\u00e1ce (de Koning a Peers 2007, Dias, Ichimura a van den Berg 2008, Forslund, Fredriksson a Vinkstr\u00f6m 2011). Richardson a van den Berg (2006) p\u0159edpokl\u00e1daj\u00ed, \u017ee skryt\u00fdm faktorem s&nbsp;vlivem na efekty programu je zv\u00fd\u0161en\u00e1 hleda\u010dsk\u00e1 aktivita zprost\u0159edkovatel\u016f zam\u011bstn\u00e1n\u00ed bezprost\u0159edn\u011b po skon\u010den\u00ed vzd\u011bl\u00e1vac\u00edho programu.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u0158e\u0161en\u00ed probl\u00e9mu potenci\u00e1ln\u00ed existence nepozorovan\u00fdch prom\u011bnn\u00fdch je obt\u00ed\u017en\u00e9. Puhani (1998) doporu\u010duje prov\u00e1d\u011bt test nepozorovan\u00e9 heterogenity. De Koning a Peers (2007) navrhuj\u00ed porovn\u00e1vat \u0161ance nezam\u011bstnan\u00fdch na nalezen\u00ed zam\u011bstn\u00e1n\u00ed p\u0159ed programem a po programu, co\u017e by mohlo pomoci, pokud se nepozorovan\u00e9 faktory nem\u011bn\u00ed v&nbsp;\u010dase \u010di r\u016fzn\u011b u&nbsp;obou srovn\u00e1van\u00fdch skupin. Lze p\u0159edpokl\u00e1dat, \u017ee zahrnut\u00ed prom\u011bnn\u00e9 historie p\u0159ed programem, kter\u00e1 indikuje \u0161ance nezam\u011bstnan\u00e9ho na trhu pr\u00e1ce, do p\u00e1rov\u00e1n\u00ed m\u016f\u017ee v\u00fdznamn\u011b pomoci sn\u00ed\u017eit riziko p\u016fsoben\u00ed nepozorovan\u00fdch faktor\u016f<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn18\"><sup>[18]<\/sup><\/a>.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Nyn\u00ed se zam\u011b\u0159\u00edme na ot\u00e1zku identifikace vhodn\u00fdch faktor\u016f pro p\u00e1rov\u00e1n\u00ed. Bryson, Dorsett a Purdon (2002) upozor\u0148uj\u00ed, \u017ee volba t\u011bchto faktor\u016f by m\u011bla b\u00fdt zalo\u017eena na dobr\u00e9 znalosti teorie a v\u00fdsledk\u016f p\u0159edchoz\u00edch v\u00fdzkum\u016f. Zpravidla jsou vyu\u017e\u00edv\u00e1ny nap\u0159. v\u011bk, pohlav\u00ed, vzd\u011bl\u00e1n\u00ed, zdravotn\u00ed stav, region \u00fa\u0159adu pr\u00e1ce, p\u0159edchoz\u00ed obor \u010dinnosti atd. Zde se kr\u00e1tce zam\u011b\u0159\u00edme na diskuzi p\u0159edev\u0161\u00edm n\u011bkter\u00fdch t\u011bchto faktor\u016f.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Jedn\u00edm z&nbsp;kl\u00ed\u010dov\u00fdch faktor\u016f \u00fasp\u011b\u0161nosti program\u016f je p\u0159ed-programov\u00e1 perspektiva \u00fa\u010dastn\u00edk\u016f programu z&nbsp;hlediska nalezen\u00ed zam\u011bstn\u00e1n\u00ed. V\u00fdzkumy (nap\u0159. Lechner a Wunsch 2009, Hora a Suchanec 2014) uk\u00e1zaly, \u017ee&nbsp;dopady program\u016f APZ jsou hor\u0161\u00ed u nezam\u011bstnan\u00fdch s&nbsp;lep\u0161\u00edmi \u0161ancemi na nalezen\u00ed zam\u011bstn\u00e1n\u00ed. Dal\u0161\u00edm v\u00fdznamn\u00fdm faktorem, kter\u00fd je nutn\u00e9 zahrnout do zji\u0161\u0165ov\u00e1n\u00ed efekt\u016f programu je p\u0159ed-programov\u00e1 historie \u00fa\u010dastn\u00edk\u016f a ne\u00fa\u010dastn\u00edk\u016f programu vzhledem k&nbsp;okolnosti, u n\u00ed\u017e o\u010dek\u00e1v\u00e1me efekt programu nap\u0159. d\u00e9lka nezam\u011bstnanosti (Bryson, Dorsett a&nbsp;Purdon 2002, Card, Ibarr\u00e1n a Villa 2011, Forslund, Fredriksson a Vinkstr\u00f6m 2011). V&nbsp;p\u0159edchoz\u00ed anal\u00fdze (Hora a Suchanec 2014) jsme prok\u00e1zali, \u017ee p\u0159edchoz\u00ed d\u00e9lka nezam\u011bstnanosti m\u00e1 v\u00fdznamn\u00fd vliv na \u0161ance nezam\u011bstnan\u00fdch odej\u00edt z&nbsp;evidence \u00da\u0159adu pr\u00e1ce. Krom\u011b t\u00e9to skute\u010dnosti maj\u00ed uveden\u00e9 informace v\u00fdznam t\u00e9\u017e pro posouzen\u00ed c\u00edlenosti programu z&nbsp;hlediska na\u010dasov\u00e1n\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V\u00fdzkumn\u00edk se m\u016f\u017ee s&nbsp;probl\u00e9mem selekce vyrovnat p\u0159edev\u0161\u00edm volbou vhodn\u00e9ho zp\u016fsobu odhadu dopadu (estim\u00e1toru) za pomoci v\u011brohodn\u00e9 identifika\u010dn\u00ed podm\u00ednky. V&nbsp;na\u0161em p\u0159\u00edpad\u011b ke&nbsp;kontra-faktu\u00e1ln\u00edmu odhadu vyu\u017e\u00edv\u00e1me p\u00e1rov\u00e1n\u00ed s pomoc\u00ed tzv. propensity score matching.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Probl\u00e9m p\u0159eb\u011bhnut\u00ed (crossover, spillover) a \u0159e\u0161en\u00ed nedokon\u010den\u00e9 \u00fa\u010dasti v&nbsp;programu<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Rozd\u011blen\u00ed nezam\u011bstnan\u00fdch na \u00fa\u010dastn\u00edky a ne\u00fa\u010dastn\u00edky je d\u00e1le komplikov\u00e1no rizikem \u0161patn\u00e9ho za\u0159azen\u00ed nezam\u011bstnan\u00fdch do skupiny intervence a&nbsp;do kontroln\u00ed skupiny (nap\u0159. p\u0159i \u00fa\u010dasti ne\u00fa\u010dastn\u00edk\u016f v&nbsp;programu APZ \u010di&nbsp;v&nbsp;jin\u00e9m programu) \u2013 tedy probl\u00e9mem tzv. \u201ecross-over\u201c (Card, Ibarr\u00e1n a&nbsp;Villa 2011).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Prvn\u00ed z&nbsp;v\u00fd\u0161e uveden\u00fdch situac\u00ed (\u0161patn\u00e9 za\u0159azen\u00ed do skupiny intervence) nast\u00e1v\u00e1, pokud panuje nejistota ohledn\u011b spr\u00e1vnosti za\u0159azen\u00ed nezam\u011bstnan\u00fdch do skupiny intervence. Card, Ibarr\u00e1n a Villa (2011) identifikuj\u00ed dv\u011b skupiny:<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>a) nezam\u011bstnan\u00e9, kte\u0159\u00ed se do programu p\u0159ihl\u00e1sili, ale pak do n\u011bj nenastoupili,<\/li>\n\n\n\n<li>b) nezam\u011bstnan\u00e9, kte\u0159\u00ed v&nbsp;programu participovali, ale program nedokon\u010dili.<\/li>\n<\/ol>\n\n\n\n<p class=\"wp-block-paragraph\">Card, Ibarr\u00e1n a Villa (2011) se domn\u00edvaj\u00ed, \u017ee ob\u011b tyto skupiny by m\u011bly b\u00fdt ch\u00e1p\u00e1ny jako \u00fa\u010dastn\u00edci skupiny intervence. \u00da\u010dast v&nbsp;programu m\u016f\u017ee b\u00fdt t\u00e9\u017e definov\u00e1na jako absolvov\u00e1n\u00ed alespo\u0148 ur\u010dit\u00e9 \u010d\u00e1sti programu (Card, Ibarr\u00e1n a Villa 2011).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">P\u0159i druh\u00e9 situaci (\u0161patn\u00e9 za\u0159azen\u00ed do kontroln\u00ed skupiny) m\u016f\u017eeme (n\u011bkdy nev\u011bdom\u011b) porovn\u00e1vat \u00fa\u010dastn\u00edky sledovan\u00e9ho programu s&nbsp;\u00fa\u010dastn\u00edky jin\u00fdch program\u016f (nap\u0159. p\u0159i chybn\u00e9m za\u0159azen\u00ed nezam\u011bstnan\u00fdch do kontroln\u00ed skupiny na z\u00e1klad\u011b chyb\u011bj\u00edc\u00ed informace o jejich \u00fa\u010dasti v&nbsp;programu). D\u00e1le Calmfors, Forslund a Hemstr\u00f6m (2002) uv\u00e1d\u00ed, \u017ee p\u0159i vy\u0161\u0161\u00edm rozsahu APZ je pravd\u011bpodobn\u00e9, \u017ee ka\u017ed\u00fd nezam\u011bstnan\u00fd se z\u00fa\u010dastnil, z\u00fa\u010dast\u0148uje nebo bude \u00fa\u010dastnit APZ, a proto je obt\u00ed\u017en\u00e9 nal\u00e9zt kontroln\u00ed skupinu bez&nbsp;jak\u00e9koliv \u00fa\u010dasti. Z&nbsp;tohoto hlediska se tak\u00e9 m\u016f\u017ee snadno st\u00e1t, \u017ee evalu\u00e1tor srovn\u00e1v\u00e1 \u00fa\u010dastn\u00edky jednoho programu s&nbsp;\u00fa\u010dastn\u00edky jin\u00e9ho programu realizovan\u00e9ho pozd\u011bji (ibid.). Tento probl\u00e9m byl pops\u00e1n v&nbsp;\u0159ad\u011b evalua\u010dn\u00edch studi\u00ed (viz nap\u0159. Heckman a Smith 1996, Sianesi 2003).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Praktick\u00fd v\u00fdznam probl\u00e9mu cross-over p\u0159i hodnocen\u00ed APZ lze identifikovat v&nbsp;n\u00e1sleduj\u00edc\u00edch situac\u00edch.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>K&nbsp;\u00fa\u010dasti nezam\u011bstnan\u00e9ho z&nbsp;kontroln\u00ed skupiny v&nbsp;programu mohlo doj\u00edt p\u0159ed programem, jeho\u017e v\u00fdsledky pl\u00e1nujeme hodnotit, v dob\u011b trv\u00e1n\u00ed tohoto programu, i dob\u011b po skon\u010den\u00ed tohoto programu.<\/li>\n\n\n\n<li>Programy mohou b\u00fdt vz\u00e1jemn\u00fdmi substituty (Heckman a Smith 1996, Richardson a van den Berg 2006).<\/li>\n\n\n\n<li>Nezam\u011bstnan\u00ed mohli b\u00fdt \u00fa\u010dastn\u00edky jin\u00fdch program\u016f APZ ji\u017e p\u0159ed&nbsp;za\u010d\u00e1tkem sledovan\u00e9ho programu co\u017e m\u016f\u017ee m\u00edt vliv jak&nbsp;na&nbsp;jejich pravd\u011bpodobnost odchodu do zam\u011bstn\u00e1n\u00ed (v p\u0159echoz\u00edm programu, po p\u0159edchoz\u00edm programu), tak na pravd\u011bpodobnost za\u0159azen\u00ed do sledovan\u00e9ho programu (Richardson a van den Berg 2006).<\/li>\n\n\n\n<li>To, \u017ee nezam\u011bstnan\u00ed z\u00edskali jin\u00fd program, m\u016f\u017ee a nemus\u00ed b\u00fdt evidov\u00e1no.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">N\u011bkte\u0159\u00ed auto\u0159i se p\u0159ikl\u00e1n\u011bj\u00ed k&nbsp;n\u00e1zoru, \u017ee nen\u00ed spr\u00e1vn\u00e9, pot\u0159ebn\u00e9 nebo nezbytn\u00e9 kl\u00e1st omezen\u00ed ohledn\u011b statusu \u00fa\u010dastn\u00edk\u016f kontroln\u00ed skupiny (Sianesi 2003, Dias, Ichimura a van den Berg 2008). Tato \u00favaha m\u016f\u017ee b\u00fdt t\u00e9\u017e modifikov\u00e1na r\u016fzn\u00fdmi \u010dasov\u00fdmi hledisky. Specifick\u00fdm \u0159e\u0161en\u00edm tohoto probl\u00e9mu je z\u00e1m\u011brn\u00e9 srovn\u00e1v\u00e1n\u00ed r\u016fzn\u00fdch skupin \u00fa\u010dastn\u00edk\u016f programu (zaj\u00edmav\u00e1 je v&nbsp;tomto ohledu nap\u0159. studie Sianesi 2003 o vlivu \u010dasov\u00e1n\u00ed vstupu do programu). Dal\u0161\u00ed mo\u017enost\u00ed je z\u00e1m\u011brn\u00e9 srovn\u00e1v\u00e1n\u00ed jednotliv\u00fdch typ\u016f program\u016f mezi sebou (Bryson, Dorsett a Purdon 2002).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u0158e\u0161en\u00ed probl\u00e9mu cross-over v&nbsp;sob\u011b zahrnuje dv\u011b roviny: (a) normativn\u00ed, tedy u\u010din\u011bn\u00ed rozhodnut\u00ed, zda osoby \u00fa\u010dastn\u00edci jin\u00fdch program\u016f zahrnovat do kontroln\u00ed skupiny, a (b) praktickou, kdy jsou auto\u0159i v&nbsp;n\u011bkter\u00fdch zem\u00edch (zejm\u00e9na ve \u0160v\u00e9dsku) k&nbsp;tomuto postupu \u201enuceni\u201c nebo\u0165 nelze nal\u00e9zt nezam\u011bstnan\u00e9 bez \u00fa\u010dasti v&nbsp;programu. V&nbsp;\u010cR je nalezen\u00ed vhodn\u00fdch nezam\u011bstnan\u00fdch do kontroln\u00ed skupiny pom\u011brn\u011b snadn\u00e9, proto se jedn\u00e1 sp\u00ed\u0161e o&nbsp;normativn\u00ed rozhodnut\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Definice \u010dasov\u00fdch bod\u016f pro m\u011b\u0159en\u00ed v\u00fdsledku a dopadu<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Z\u00e1kladem pro vymezen\u00ed \u010dasov\u00fdch bod\u016f je vymezen\u00ed sledovan\u00e9ho obdob\u00ed. Jedn\u00e1 se o takov\u00e9 obdob\u00ed, ke kter\u00e9mu lze logicky vztahovat prezentovan\u00e1 data a kter\u00e9 odpov\u00edd\u00e1 z\u00e1m\u011br\u016fm hodnocen\u00ed. P\u0159edev\u0161\u00edm se jedn\u00e1 o:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>informace o nezam\u011bstnan\u00fdch (\u00fa\u010dastn\u00edc\u00edch a ne\u00fa\u010dastn\u00edc\u00edch program\u016f),<\/li>\n\n\n\n<li>informace o programech,<\/li>\n\n\n\n<li>informace o evidenc\u00edch.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">V&nbsp;r\u00e1mci jedn\u00e9 evaluace je mo\u017en\u00e9 m\u00edt sledovan\u00e9 obdob\u00ed definov\u00e1no r\u016fzn\u011b pro r\u016fzn\u00e9 okruhy \u00fadaj\u016f (nem\u011bnn\u00e9 \u00fadaje nap\u0159. pohlav\u00ed se zpravidla sleduj\u00ed pouze jednou).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V\u00fdznamn\u00fdm metodologick\u00fdm probl\u00e9mem je pr\u00e1ce s&nbsp;\u00fadaji, kter\u00e9 jsou u\u017eite\u010dn\u00e9 pro hodnocen\u00ed, ale nejsou \u010d\u00e1ste\u010dn\u011b dostupn\u00e9 nebo se nach\u00e1zej\u00ed vn\u011b sledovan\u00e9ho obdob\u00ed (tzv. cenzory). Jedn\u00e1 se tedy o formu \u010d\u00e1ste\u010dn\u011b chyb\u011bj\u00edc\u00edch \u00fadaj\u016f (viz Vermunt 1996). D\u016fvodem existence cenzor\u016f je neznalost historie p\u0159ed sledovan\u00fdm obdob\u00edm (lev\u00fd cenzor), kone\u010dnost sledovan\u00e9ho obdob\u00ed (ty, kdo p\u0159i\u0161li pozd\u011bji, je mo\u017en\u00e9 sledovat po krat\u0161\u00ed dobu \u2013 prav\u00fd cenzor) a nast\u00e1n\u00ed jin\u00e9 ud\u00e1losti \u010di zmizen\u00ed z&nbsp;datov\u00e9ho souboru (nev\u00edme, co&nbsp;se s&nbsp;osobami stalo) b\u011bhem sledovan\u00e9ho obdob\u00ed (Vermunt 1996).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Probl\u00e9m nast\u00e1v\u00e1 p\u0159edev\u0161\u00edm tehdy, pokud m\u016f\u017eeme o\u010dek\u00e1vat, \u017ee chyb\u011bj\u00edc\u00ed \u00fadaje mohou zkreslovat v\u00fdsledky program\u016f. Vermunt (1996) uv\u00e1d\u00ed, \u017ee&nbsp;vy\u0159azen\u00ed p\u0159\u00edpad\u016f s&nbsp;chyb\u011bj\u00edc\u00edmi (cenzorovan\u00fdmi) hodnotami m\u016f\u017ee z\u00e1va\u017en\u011b zkreslit v\u00fdsledky, pokud nen\u00ed distribuce t\u011bchto p\u0159\u00edpad\u016f n\u00e1hodn\u00e1, a&nbsp;proto doporu\u010duje zachovat maxim\u00e1ln\u00ed mo\u017en\u00e9 mno\u017estv\u00ed informac\u00ed a sna\u017eit se zabr\u00e1nit zkreslen\u00ed selekce. K&nbsp;\u0159e\u0161en\u00ed cenzor\u016f zprava jsou vyu\u017e\u00edv\u00e1ny techniky vych\u00e1zej\u00edc\u00ed z&nbsp;p\u0159edpokladu, \u017ee m\u00edra cenzorov\u00e1n\u00ed je nez\u00e1visl\u00e1 na&nbsp;sledovan\u00e9m v\u00fdsledku. Jedn\u00edm z&nbsp;\u0159e\u0161en\u00ed je kontrolovat pro data vstupu do anal\u00fdzy a dal\u0161\u00ed prom\u011bnn\u00e9, kter\u00e9 by mohly b\u00fdt p\u0159\u00ed\u010dinou systematick\u00e9ho cenzorov\u00e1n\u00ed (viz ibid.). Vermunt (1996) d\u00e1le p\u00ed\u0161e, \u017ee v p\u0159\u00edpad\u011b existence cenzor\u016f zleva je mo\u017en\u00e9 situaci \u0159e\u0161it odstran\u011bn\u00edm t\u011bchto p\u0159\u00edpad\u016f (doporu\u010duje se, jen pokud jich je jen mal\u00fd po\u010det), zat\u00edmco odhadov\u00e1n\u00ed cenzorovan\u00fdch dat je obt\u00ed\u017en\u00e9. Je toti\u017e mo\u017en\u00e9 p\u0159edpokl\u00e1dat, \u017ee p\u0159\u00edpady cenzorovan\u00e9 zleva neodpov\u00eddaj\u00ed p\u0159\u00edpad\u016fm, o kter\u00fdch m\u00e1me kompletn\u00ed informace (doch\u00e1z\u00ed ke zkreslen\u00ed selekce). B\u011b\u017en\u00fdm \u0159e\u0161en\u00edm lev\u00fdch cenzor\u016f je tedy vy\u0159azen\u00ed t\u011bchto p\u0159\u00edpad\u016f z&nbsp;anal\u00fdzy (nap\u0159. Puhani 1998). V&nbsp;p\u0159\u00edpad\u011b dat OKpr\u00e1ce nen\u00ed probl\u00e9mem p\u0159\u00edtomnost lev\u00fdch cenzor\u016f, zat\u00edmco p\u0159\u00edtomnost prav\u00fdch cenzor\u016f je z&nbsp;logiky v\u011bci \u010dasto nevyhnuteln\u00e1.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Evaluace vy\u017eaduje definice n\u011bkter\u00fdch p\u0159esn\u00fdch \u010dasov\u00fdch bod\u016f. Podle&nbsp;konkr\u00e9tn\u00edho typu hodnocen\u00ed pot\u0159ebujeme identifikovat situaci p\u0159ed&nbsp;programem, v&nbsp;programu a po programu.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Pro identifikaci p\u0159ed-programov\u00e9ho stavu je podle Card, Ibarr\u00e1n a Villa (2011) jako z\u00e1kladn\u00ed bod sledov\u00e1na situace t\u011bsn\u011b p\u0159ed za\u010d\u00e1tkem programu. \u0158ada autor\u016f upozor\u0148uje, \u017ee situace t\u011bsn\u011b p\u0159ed programem se m\u016f\u017ee v\u00fdrazn\u011b odli\u0161ovat od d\u0159\u00edv\u011bj\u0161\u00ed situace nezam\u011bstnan\u00fdch. Hujer, Thomsen a&nbsp;Zeiss (2006) nap\u0159. p\u0159edpokl\u00e1daj\u00ed, \u017ee nezam\u011bstnan\u00ed, kte\u0159\u00ed se dozv\u011bd\u00ed o sv\u00e9 \u00fa\u010dasti v&nbsp;programu, mohou zm\u011bnit sv\u00e9 strategie hled\u00e1n\u00ed zam\u011bstn\u00e1n\u00ed. Vliv tohoto efektu (popsan\u00e9ho poprv\u00e9 Ashelferterem) se pravd\u011bpodobn\u011b bude odli\u0161ovat t\u00e9\u017e podle zvolen\u00e9ho v\u00fdsledku. V&nbsp;p\u0159\u00edpad\u011b vyu\u017eit\u00ed dat OKpr\u00e1ce a&nbsp;volby v\u00fdsledku (ne)p\u0159\u00edtomnost v&nbsp;evidenci jsou zpravidla p\u0159ed programem v\u0161ichni \u00fa\u010dastn\u00edc\u00ed programu ch\u00e1p\u00e1ni jako nezam\u011bstnan\u00ed (implicitn\u00ed pre-test) a sledov\u00e1n je jen pod\u00edl v evidenci\/zam\u011bstnan\u00fdch po programu.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Velmi v\u00fdznamnou ot\u00e1zkou je, zda m\u011b\u0159it dopady program\u016f od za\u010d\u00e1tku nebo od konce programu, a jak v&nbsp;takov\u00e9m p\u0159\u00edpad\u011b definovat \u010dasov\u00e9 body pro za\u010d\u00e1tek sledov\u00e1n\u00ed u kontroln\u00ed skupiny. N\u011bkter\u00e9 studie prok\u00e1zaly, \u017ee&nbsp;efekty program\u016f se mohou odli\u0161ovat v&nbsp;z\u00e1vislosti na zvolen\u00ed jednoho nebo druh\u00e9ho z&nbsp;t\u011bchto p\u0159\u00edstup\u016f (srovnej Calmfors, Forslund a Hemstr\u00f6m 2002).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">U skupiny intervence m\u016f\u017eeme uva\u017eovat o n\u00e1sleduj\u00edc\u00edch \u010dasov\u00fdch bodech: a) za\u010d\u00e1tek evidence p\u0159ed programem (p\u0159\u00edpadn\u011b dal\u0161\u00ed p\u0159ed-programov\u00e9 evidence), b) datum za\u010d\u00e1tku programu (data za\u010d\u00e1tku v\u00edce program\u016f), c)&nbsp;datum konce programu (konc\u016f v\u00edce program\u016f).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Pro odvozen\u00ed z\u00e1kladn\u00edho bodu v\u00fdpo\u010dtu pro kontroln\u00ed skupinu jsou vyu\u017e\u00edv\u00e1ny r\u016fzn\u00e9 p\u0159\u00edstupy (viz nap\u0159. Larson 2001 podle Calmfors, Forslund a&nbsp;Hemstr\u00f6m 2002, Sianesi 2003, Lechner a Wunsch 2009).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Larson (2001 podle Calmfors, Forslund a Hemstr\u00f6m 2002) uv\u00e1d\u00ed, \u017ee pokud sledujeme v\u00fdsledky programu po programu, a ne od po\u010d\u00e1tku programu, m\u016f\u017eeme stanovit startovn\u00ed bod pro kontroln\u00ed skupinu dv\u011bma zp\u016fsoby:<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>a) za\u0159azen\u00edm \u010dlen\u016f kontroln\u00ed skupiny, kte\u0159\u00ed se stali nezam\u011bstnan\u00fdmi ve stejn\u00e9m \u010dase jako \u00fa\u010dastn\u00edci skupiny intervence a v&nbsp;dob\u011b ukon\u010den\u00ed programu byli st\u00e1le nezam\u011bstnan\u00ed,<\/li>\n\n\n\n<li>b) za\u0159azen\u00edm \u010dlen\u016f kontroln\u00ed skupiny, kte\u0159\u00ed byli v&nbsp;dob\u011b ukon\u010den\u00ed programu stejn\u011b dlouho nezam\u011bstnan\u00ed jako \u00fa\u010dastn\u00edci skupiny intervence p\u0159ed programem.<\/li>\n<\/ol>\n\n\n\n<p class=\"wp-block-paragraph\">Podle Larsona oba tyto zp\u016fsoby vedou ke zkreslen\u00ed (zv\u00fd\u0161en\u00ed) dopadu programu. V&nbsp;prvn\u00edm p\u0159\u00edpad\u011b proto, \u017ee v&nbsp;kontroln\u00ed skupin\u011b budou nadm\u011brn\u011b zastoupeni nezam\u011bstnan\u00ed s&nbsp;hor\u0161\u00edmi \u0161ancemi na trhu pr\u00e1ce. Ve druh\u00e9m p\u0159\u00edpad\u011b z\u00edsk\u00e1vaj\u00ed \u00fa\u010dastn\u00edci programu del\u0161\u00ed \u010das na nalezen\u00ed zam\u011bstn\u00e1n\u00ed, by\u0165 lze o\u010dek\u00e1vat men\u0161\u00ed intenzitu jejich hled\u00e1n\u00ed v&nbsp;pr\u016fb\u011bhu realizace programu.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Pravd\u011bpodobn\u011b z&nbsp;t\u011bchto d\u016fvod\u016f Puhani (1998) a Forslund, Fredriksson a&nbsp;Vinkstr\u00f6m (2011) sleduj\u00ed v\u00fdsledky programu od po\u010d\u00e1tku programu. N\u011bkte\u0159\u00ed auto\u0159i d\u00e1le rozli\u0161uj\u00ed efekt intervence na \u010d\u00e1st b\u011bhem programu a&nbsp;na&nbsp;\u010d\u00e1st po programu. Puhani (1998), Sianesi (2003) a Hora a Suchanec (2014) stanovuj\u00ed podm\u00ednku, \u017ee jedinci v&nbsp;kontroln\u00ed skupin\u011b str\u00e1vili v nezam\u011bstnanosti alespo\u0148 takovou dobu, jako \u00fa\u010dastn\u00edci programu p\u0159ed vstupem do programu. C\u00edlem tohoto p\u0159\u00edstupu je postavit \u00fa\u010dastn\u00edky na stejnou \u201estartovn\u00ed \u010d\u00e1ru\u201c. Puhani (1998) tento postup zd\u016fvod\u0148uje t\u00edm, \u017ee historie p\u0159ed programem je zn\u00e1m\u00e1 pouze pro skupinu intervence, a proto nem\u016f\u017ee b\u00fdt zahrnuta do propensity score.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Z\u00e1sadn\u00edm poznatkem z&nbsp;hlediska hodnocen\u00ed dopad\u016f APZ je mo\u017enost prom\u011bnlivosti dopadu APZ v&nbsp;\u010dase, kter\u00e1 m\u016f\u017ee nastat nap\u0159. tehdy, pokud je k&nbsp;projeven\u00ed efektu pot\u0159eba ur\u010dit\u00fd \u010das \u010di naopak, pokud k&nbsp;efektu programu doch\u00e1z\u00ed nejv\u00edce bezprost\u0159edn\u011b po skon\u010den\u00ed programu (Richardson a&nbsp;van den Berg 2006, Hujer, Thomsen a Zeiss 2006). Evalua\u010dn\u00ed literatura zpravidla rozli\u0161uje kr\u00e1tkodob\u00e9, st\u0159edn\u011bdob\u00e9 a dlouhodob\u00e9 v\u00fdsledky program\u016f. Jako kr\u00e1tkodob\u00e9 v\u00fdsledky jsou ozna\u010dov\u00e1ny v\u00fdsledky do jednoho roku, jako st\u0159edn\u011bdob\u00e9 a dlouhodob\u00e9 v\u00fdsledky pak v\u00fdsledky v&nbsp;obdob\u00ed v\u00edce ne\u017e dvou let (viz nap\u0159. Card, Kluve a Weber 2009, Card, Ibarr\u00e1n a&nbsp;Villa 2011)<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn19\"><sup>[19]<\/sup><\/a>. Prezentace dlouhodob\u00fdch v\u00fdsledk\u016f je \u010dasto v&nbsp;\u010cR i v zahrani\u010d\u00ed omezena nedostupnost\u00ed vhodn\u00fdch dat a snahou p\u0159in\u00e9st aktu\u00e1ln\u00ed \u00fadaje (viz nap\u0159. Lechner a Wunsch 2009). V&nbsp;modelech sleduj\u00edc\u00edch riziko sp\u00ed\u0161e hovo\u0159\u00edme o osob\u00e1ch vystaven\u00fdch riziku, o rizikov\u00e9m obdob\u00ed, tedy obdob\u00ed, kdy je osoba vystavena riziku a trv\u00e1n\u00ed, ne\u017e do\u0161lo ke sledovan\u00e9 ud\u00e1losti (viz Vermunt 1996).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Diskuze p\u016fsoben\u00ed \u010dasu a jeho vlivu na evaluaci program\u016f APZ<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V\u00fdsledky program\u016f APZ mohou b\u00fdt ovlivn\u011bny cyklick\u00fdmi a sez\u00f3nn\u00edmi vlivy (Card, Ibarr\u00e1n a Villa 2011, Forslund, Fredriksson a Vinkstr\u00f6m 2011). Rozsah program\u016f APZ m\u00e1 \u010dasto pro-cyklick\u00fd charakter (viz nap\u0159. Calmfors, Forslund a Hemstr\u00f6m 2002). Forslund, Fredriksson a Vinkstr\u00f6m (2011) p\u0159edpokl\u00e1daj\u00ed (a empiricky potvrdili), \u017ee hlavn\u00edm faktorem je zde rozd\u00edln\u00e1 \u0161ance nezam\u011bstnan\u00fdch nal\u00e9zt si zam\u011bstn\u00e1n\u00ed v&nbsp;konjunktu\u0159e a&nbsp;v&nbsp;recesi, kter\u00e1 je v&nbsp;interakci s&nbsp;participac\u00ed v&nbsp;programu. Srovn\u00e1n\u00ed efekt\u016f v&nbsp;r\u016fzn\u00fdch \u010d\u00e1stech hospod\u00e1\u0159sk\u00e9ho cyklu vy\u017eaduje napln\u011bn\u00ed \u0159ady dal\u0161\u00edch p\u0159edpoklad\u016f o shodnosti dal\u0161\u00edch okolnost\u00ed ve srovn\u00e1van\u00fdch obdob\u00edch (viz&nbsp;Forslund, Fredriksson a Vinkstr\u00f6m 2011).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Kluve et al. (2005) a Borland, Tseng a Wilkins (2005) uv\u00e1d\u011bj\u00ed, \u017ee kvazi-experiment\u00e1ln\u00ed postupy jsou z\u00e1visl\u00e9 na p\u0159edpokladu stabiln\u00ed hodnoty intervence (SUTVA). Tento p\u0159edpoklad zahrnuje o\u010dek\u00e1v\u00e1n\u00ed, \u017ee: (a)&nbsp;efekt intervence na ka\u017ed\u00e9ho jednotlivce nen\u00ed ovlivn\u011bn rozhodnut\u00edm o participaci dal\u0161\u00edch jednotlivc\u016f (Bryson, Dorsett a Purdon 2002, Kluve et&nbsp;al. 2005) \u010di&nbsp;tak\u00e9 (b) efekt \u00fa\u010dasti v&nbsp;programu je v&nbsp;\u010dase stabiln\u00ed a&nbsp;(c)&nbsp;v\u00fdsledky ne\u00fa\u010dastn\u00edk\u016f nejsou programem ovlivn\u011bny (Bryson, Dorsett a Purdon 2002, Borland, Tseng a Wilkins 2005). Richardson a van den Berg (2006) uv\u00e1d\u011bj\u00ed, \u017ee&nbsp;m\u00edra odchodu z&nbsp;evidence po programu se v&nbsp;\u010dase odli\u0161uje (kles\u00e1). Auto\u0159i upozor\u0148uj\u00ed, \u017ee tento jev m\u016f\u017ee b\u00fdt spojen jak s&nbsp;klesaj\u00edc\u00edm efektem programu, tak s&nbsp;vlivem nepozorovan\u00fdch prom\u011bnn\u00fdch.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Richardson a van den Berg (2006) nap\u0159. uv\u00e1d\u011bj\u00ed, \u017ee jejich model vych\u00e1z\u00ed z&nbsp;p\u0159edpokladu neexistence vlivu programu p\u0159ed programem. K&nbsp;takov\u00e9mu vlivu m\u016f\u017ee doch\u00e1zet, pokud kandid\u00e1ti \u00fa\u010dasti v&nbsp;programu anticipuj\u00ed svou \u00fa\u010dast a m\u011bn\u00ed na z\u00e1klad\u011b toho sv\u00e9 chov\u00e1n\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Techniky vyhodnocen\u00ed v\u00fdsledku<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Z&nbsp;hlediska pojet\u00ed \u010dasu m\u016f\u017eeme v&nbsp;z\u00e1vislosti na povaze dat a metodologick\u00fdch z\u00e1m\u011brech evalu\u00e1tor\u016f rozli\u0161it p\u0159\u00edstupy zalo\u017een\u00e9 na nespojit\u00e9m \u010dase (discrete time), ud\u00e1lost m\u016f\u017ee nastat v&nbsp;ur\u010dit\u00fd ur\u010den\u00fd \u010das, a p\u0159\u00edstupy zalo\u017een\u00e9 na spojit\u00e9m \u010dase (continuous time), kde ud\u00e1lost m\u016f\u017ee nastat kdykoliv (Vermunt 1996). Nezam\u011bstnan\u00fd m\u016f\u017ee m\u00edt b\u011bhem sledovan\u00e9ho obdob\u00ed jednu \u010di v\u00edce evidenc\u00ed, kter\u00e9 jsou ohrani\u010deny p\u0159esn\u00fdmi \u010dasov\u00fdmi body. Data OKpr\u00e1ce maj\u00ed kl\u00ed\u010dov\u00e9 \u00fadaje uvedeny s&nbsp;p\u0159esnost\u00ed na jednotliv\u00e9 dny, a&nbsp;proto umo\u017e\u0148uj\u00ed vyu\u017eit\u00ed ob\u011bma z\u00e1kladn\u00edmi zp\u016fsoby. V&nbsp;evalua\u010dn\u00ed literatu\u0159e (de Koning a Peers 2007, Card, Kluve a Weber 2009) se d\u00e1le setk\u00e1v\u00e1me s&nbsp;n\u00e1sleduj\u00edc\u00edmi dv\u011bma p\u0159\u00edstupy k&nbsp;hodnocen\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">V&nbsp;prvn\u00edm p\u0159\u00edpad\u011b je v\u00fdsledek definov\u00e1n jako stav v&nbsp;ur\u010dit\u00e9m \u010dasov\u00e9m bod\u011b, tedy nap\u0159. kolik osob nebylo v&nbsp;ur\u010dit\u00e9m \u010dasov\u00e9m bod\u011b v evidenci. A\u010dkoliv je tento v\u00fdsledek n\u011bkdy t\u00e9\u017e interpretov\u00e1n jako \u201ekolik nezam\u011bstnan\u00fdch do ur\u010dit\u00e9 doby opustilo evidenci \u00da\u0159adu pr\u00e1ce\u201c, je t\u0159eba vz\u00edt v&nbsp;\u00favahu, \u017ee z\u00e1le\u017e\u00ed na tom, zda sledujeme situaci i v&nbsp;jin\u00fdch \u010dasov\u00fdch bodech. Vermunt (1996) upozor\u0148uje, \u017ee zm\u011bnu zjist\u00edme v&nbsp;tomto p\u0159\u00edpad\u011b pouze tehdy, pokud se ob\u011b m\u011b\u0159en\u00ed odli\u0161uj\u00ed. Na druh\u00e9 stran\u011b, pokud se ob\u011b hodnoty shoduj\u00ed, mohlo \u010di nemuselo mezi jednotliv\u00fdmi m\u011b\u0159en\u00edmi doj\u00edt ke&nbsp;zm\u011bn\u011b. Nap\u0159. osoba, kter\u00e1 je v&nbsp;evidenci po roce od ukon\u010den\u00ed programu mohla b\u00fdt b\u011bhem tohoto roku 11 m\u011bs\u00edc\u016f mimo evidenci, a p\u0159esto bychom mohli myln\u011b p\u0159edpokl\u00e1dat, \u017ee byla v&nbsp;evidenci \u00daP po cel\u00fdch 12 m\u011bs\u00edc\u016f. Tot\u00e9\u017e plat\u00ed pro obdob\u00ed n\u00e1sleduj\u00edc\u00ed po sledovan\u00e9m \u010dasov\u00e9m bod\u011b. Je ov\u0161em mo\u017en\u00e9 sledovat situaci ve v\u00edce \u010dasov\u00fdch bodech. Pro tento typ anal\u00fdzy je mo\u017en\u00e9 vyu\u017e\u00edt nap\u0159. logistickou regresi (Bryson, Dorsett a Purdon 2002). P\u0159\u00edkladem evaluace proveden\u00e9 t\u00edmto zp\u016fsobem je nap\u0159. Reinowski a&nbsp;Schultz (2006).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Ve druh\u00e9m p\u0159\u00edpad\u011b hovo\u0159\u00edme o \u010dasov\u00e1n\u00ed ud\u00e1lost\u00ed (timing of events) ve&nbsp;kter\u00e9m jsou vyu\u017e\u00edv\u00e1na srovn\u00e1v\u00e1n\u00ed rizik na odchod z&nbsp;evidence v&nbsp;r\u016fzn\u00fdch \u010dasov\u00fdch obdob\u00edch nap\u0159. p\u0159ed programem a po programu (Richardson a van den Berg 2006, de Koning a Peers 2007). V&nbsp;t\u011bchto modelech je sledov\u00e1no, zda ud\u00e1losti nastaly, p\u0159\u00edpadn\u011b v&nbsp;jak\u00e9m po\u0159ad\u00ed, z\u00e1kladn\u00ed jednotkou pro zji\u0161\u0165ov\u00e1n\u00ed dopadu je trv\u00e1n\u00ed \u010dasu, ne\u017e poprv\u00e9 \u010di opakovan\u011b nastane \u010di nenastane sledovan\u00e1 ud\u00e1lost (Vermunt 1996). Vermunt (1996) rozd\u011bluje modely na parametrick\u00e9 modely zalo\u017een\u00e9 na spojit\u00e9m \u010dase (kam za\u0159azuje anal\u00fdzu p\u0159e\u017eit\u00ed<a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftn20\"><sup>[20]<\/sup><\/a>), semi-parametrick\u00e9 modely zalo\u017een\u00e9 na spojit\u00e9m \u010dase (Coxova regrese) a anal\u00fdzy zalo\u017een\u00e9 na nespojit\u00e9m \u010dase (p\u0159edev\u0161\u00edm r\u016fzn\u00e9 varianty \u201ediscrete time hazard rate model\u201c). Modely p\u0159edpokl\u00e1daj\u00ed r\u016fznou (nikoliv stejnou) d\u00e9lku individu\u00e1ln\u00ed \u00fa\u010dasti v nezam\u011bstnanosti p\u0159ed programem (Richardson a van den Berg 2006). N\u011bkter\u00e9 modely umo\u017e\u0148uj\u00ed \u010dasov\u011b se odli\u0161uj\u00edc\u00ed hodnoty nez\u00e1visl\u00fdch prom\u011bnn\u00fdch (viz Vermunt 1996). V takov\u00e9m p\u0159\u00edpad\u011b tedy m\u016f\u017eeme nap\u0159. zachytit, \u017ee se b\u011bhem sledovan\u00e9ho obdob\u00ed zm\u011bnila \u00farove\u0148 form\u00e1ln\u00edho vzd\u011bl\u00e1n\u00ed \u00fa\u010dastn\u00edka programu. Pr\u00e1ce s&nbsp;\u010dasov\u011b prom\u011bnliv\u00fdmi hodnotami nez\u00e1visl\u00fdch prom\u011bnn\u00fdch m\u016f\u017ee b\u00fdt v\u00fdznamn\u00e1 p\u0159edev\u0161\u00edm tam, kde existuje teoretick\u00e9 opodstatn\u011bn\u00ed pro tento p\u0159\u00edstup. P\u0159\u00edklady evaluac\u00ed proveden\u00fdch t\u00edmto zp\u016fsobem jsou nap\u0159. van Ours (2002) \u010di Richardson a van den Berg (2006).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">P\u0159i rozhodov\u00e1n\u00ed o volb\u011b technik anal\u00fdzy zva\u017eujeme zejm\u00e9na odpov\u011bdi na&nbsp;\u010dty\u0159i ot\u00e1zky souvisej\u00edc\u00ed s&nbsp;vyty\u010den\u00fdm c\u00edlem evaluace:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Chceme zn\u00e1t pr\u016fb\u011bh velikosti rizika odchodu z&nbsp;nezam\u011bstnanosti v&nbsp;\u010dase nebo si vysta\u010d\u00edme s&nbsp;velikost\u00ed rizika v&nbsp;ur\u010dit\u00e9m \u010dasov\u00e9m bod\u011b (nap\u0159. p\u016fl roku po programu)?<\/li>\n\n\n\n<li>Chceme zn\u00e1t velikost rizika odchodu z&nbsp;registru v&nbsp;ur\u010dit\u00e9m \u010dase<br>nebo informaci o tom, jak\u00e1 \u010d\u00e1st nezam\u011bstnan\u00fdch ji\u017e registr opustila?<\/li>\n\n\n\n<li>Chceme zn\u00e1t velikost skute\u010dn\u00e9ho (absolutn\u00edho) rizika odchodu nebo se spokoj\u00edme s&nbsp;velikost\u00ed rizika ve srovn\u00e1n\u00ed s&nbsp;jinou skupinou nezam\u011bstnan\u00fdch?<\/li>\n\n\n\n<li>Po\u017eadujeme dodate\u010dnou kontrolu efekt\u016f jin\u00fdch prom\u011bnn\u00fdch?<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">Modelov\u00e1n\u00ed pomoc\u00ed logistick\u00e9 regrese neumo\u017e\u0148uje posouzen\u00ed rizika a t\u00edm i dopadu programu v pr\u016fb\u011bhu \u010dasu, n\u00fdbr\u017e je zapot\u0159eb\u00ed, aby evalu\u00e1tor zakotvil svou anal\u00fdzu v&nbsp;ur\u010dit\u00e9m v\u00fdb\u011brov\u00e9m okam\u017eiku (\u010dasto po jednotliv\u00fdch m\u011bs\u00edc\u00edch). Logistick\u00e1 regrese tak\u00e9 volbou ur\u010dit\u00e9ho \u010dasov\u00e9ho bodu neumo\u017e\u0148uje posouzen\u00ed rizika odchodu, ale pouze posuzuje p\u0159\u00edtomnost \u010di&nbsp;nep\u0159\u00edtomnost nezam\u011bstnan\u00fdch v&nbsp;registru. Silnou str\u00e1nkou je mo\u017enost posouzen\u00ed absolutn\u00edch rizik opu\u0161t\u011bn\u00ed rizik. Slabinou je naopak mo\u017enost zkreslen\u00ed dopadu nejen z&nbsp;d\u016fvodu zakotvenosti anal\u00fdzy pouze ve sv\u00e9voln\u011b vybran\u00fdch \u010dasov\u00fdch bodech, ale i omezen\u00e1 mo\u017enost vypo\u0159\u00e1dat se ve sledovan\u00e9m obdob\u00ed s chyb\u011bj\u00edc\u00edmi informacemi o odchodu nezam\u011bstnan\u00fdch (tzv. cenzory).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Modelov\u00e1n\u00ed pomoc\u00ed anal\u00fdzy p\u0159e\u017eit\u00ed poskytuje z\u0159ejm\u011b nejp\u0159ehledn\u011bj\u0161\u00ed obr\u00e1zek o pr\u016fb\u011bhu rizik odchodu nezam\u011bstnan\u00fdch z&nbsp;registru (a srovn\u00e1n\u00edm t\u011bchto rizik u skupiny \u00fa\u010dastn\u00edk\u016f a ne\u00fa\u010dastn\u00edk\u016f i o dopadu programu) v&nbsp;\u010dase. Nav\u00edc umo\u017e\u0148uje posouzen\u00ed absolutn\u00edch rizik v&nbsp;\u010dase i zji\u0161t\u011bn\u00ed velikosti proporce \u00fa\u010dastn\u00edk\u016f, kte\u0159\u00ed ji\u017e registr opustili v&nbsp;jak\u00e9mkoli \u010dasov\u00e9m bod\u011b. Kontrola efektu jin\u00fdch prom\u011bnn\u00fdch je mo\u017en\u00e1 skrze zastoupen\u00ed t\u011bchto prom\u011bnn\u00fdch ve zkouman\u00fdch kombinac\u00edch, nicm\u00e9n\u011b p\u0159i kombinac\u00edch vy\u0161\u0161\u00edho \u0159\u00e1du velmi pracn\u00e1. Metoda je tak\u00e9 vhodn\u00e1 v&nbsp;p\u0159\u00edpad\u011b, kdy chceme zn\u00e1t pr\u016fb\u011bh odchodu z&nbsp;registru (nebo i pr\u016fb\u011bh velikosti dopadu programu) u konkr\u00e9tn\u00ed podskupiny zam\u011bstnan\u00fdch za konkr\u00e9tn\u00edch podm\u00ednek.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Modelov\u00e1n\u00ed dopadu pomoc\u00ed Coxov\u00fdch model\u016f proporcion\u00e1ln\u00edch rizik p\u0159in\u00e1\u0161\u00ed v\u00fdhodu snadn\u00e9 kontroly efektu ru\u0161iv\u00fdch prom\u011bnn\u00fdch, nicm\u00e9n\u011b m\u00e1 \u0159adu nev\u00fdhod spo\u010d\u00edvaj\u00edc\u00edch zejm\u00e9na v&nbsp;zach\u00e1zen\u00ed s&nbsp;\u010dasem. V&nbsp;z\u00e1kladn\u00ed podob\u011b Cox\u016fv model informaci o \u010dase nepod\u00e1v\u00e1 \u2013 efekty jsou v&nbsp;\u010dase \u201ezpr\u016fm\u011brov\u00e1ny\u201c a rozd\u00edly v&nbsp;rizic\u00edch mezi posuzovan\u00fdmi skupinami nezam\u011bstnan\u00fdch jsou tak proporcion\u00e1ln\u011b stejn\u00e9 bez ohledu na b\u011bh \u010dasu. P\u0159edpoklad proporcionality je mo\u017en\u00e9 uvolnit roz\u0161\u00ed\u0159en\u00edm modelu o line\u00e1rn\u00ed interakci velikosti dopadu s&nbsp;\u010dasem, nicm\u00e9n\u011b nar\u016fst\u00e1 tak pracnost a d\u00edky line\u00e1rn\u00edmu p\u0159edpokladu nez\u00edsk\u00e1v\u00e1me re\u00e1ln\u00fd pr\u016fb\u011bh dopadu. Druh\u00e1 slabina Coxova modelu spo\u010d\u00edv\u00e1 v&nbsp;proporcion\u00e1ln\u00edm a nikoli absolutn\u00edm zhodnocen\u00ed rizik \u2013 velikost rizik ur\u010dit\u00e9 skupiny je v\u017edy posuzov\u00e1na relativn\u011b vzhledem k&nbsp;jin\u00e9 skupin\u011b, p\u0159i\u010dem\u017e mo\u017enost zhodnotit re\u00e1lnou velikost rizika odchodu ur\u010dit\u00e9 skupiny v&nbsp;ur\u010dit\u00e9m \u010dasov\u00e9m bod\u011b je nemo\u017en\u00e1.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Vypo\u0159\u00e1d\u00e1n\u00ed se s&nbsp;t\u011bmito probl\u00e9my nab\u00edz\u00ed metoda modelov\u00e1n\u00ed rizik v diskr\u00e9tn\u00edm \u010dase, kter\u00e1 umo\u017e\u0148uje uvolnit jak p\u0159edpoklad proporcionality rizik, tak p\u0159edpoklad line\u00e1rn\u00ed zm\u011bny dopadu v&nbsp;\u010dase. V\u00fdm\u011bnou za to v\u0161ak umo\u017e\u0148uje posouzen\u00ed zm\u011bny dopadu v&nbsp;\u010dase pouze v&nbsp;ur\u010dit\u00fdch \u010dasov\u00fdch bodech, nap\u0159. po m\u011bs\u00edc\u00edch. Je tedy z\u0159ejm\u00e9, \u017ee ka\u017ed\u00e1 metoda m\u00e1 sv\u00e9 siln\u00e9 a&nbsp;slab\u00e9 str\u00e1nky a v\u00fdb\u011br metody z\u00e1vis\u00ed na evalu\u00e1torov\u00fdch odpov\u011bd\u00edch na&nbsp;v\u00fd\u0161e uveden\u00e9 ot\u00e1zky.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Z\u00e1v\u011br<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">C\u00edlem t\u00e9to studie bylo definovat a diskutovat z\u00e1kladn\u00ed oblasti, principy a&nbsp;postupy hodnocen\u00ed v\u00fdsledk\u016f a dopad\u016f program\u016f aktivn\u00ed politiky zam\u011bstnanosti s&nbsp;vyu\u017eit\u00edm administrativn\u00edch dat syst\u00e9mu OKpr\u00e1ce. D\u016fraz byl kladen zejm\u00e9na na diskuzi postup\u016f tak, aby v\u00fdsledkem jejich pou\u017eit\u00ed byl validn\u00ed indik\u00e1tor dopadu. Zab\u00fdvali jsme se zejm\u00e9na intern\u00ed validitou nam\u011b\u0159en\u00e9ho dopadu, tedy do jak\u00e9 m\u00edry r\u016fzn\u00e9 postupy p\u0159i evaluaci p\u0159in\u00e1\u0161ej\u00ed v\u00fdsledek, je\u017e je prost\u00fd r\u016fzn\u00fdch zdroj\u016f zkreslen\u00ed, a utv\u00e1\u0159\u00ed tak adekv\u00e1tn\u00ed p\u0159edstavu o dopadu programu. Jako prost\u0159edek pro dodate\u010dn\u00e9 vybalancov\u00e1n\u00ed p\u0159ed-programov\u00fdch rozd\u00edl\u016f mezi skupinou intervence a kontroln\u00ed skupinou, a tud\u00ed\u017e i zp\u016fsob zv\u00fd\u0161en\u00ed validity diskutujeme proces p\u00e1rov\u00e1n\u00ed, p\u0159i\u010dem\u017e zmi\u0148ujeme d\u016fle\u017eitost zahrnut\u00ed faktor\u016f p\u0159ed-programov\u00e1 perspektiva a historie nezam\u011bstnan\u00fdch na trhu pr\u00e1ce a f\u00e1ze hospod\u00e1\u0159sk\u00e9ho cyklu do tohoto procesu. V&nbsp;posledn\u00ed \u010d\u00e1sti jsme diskutovali volbu konkr\u00e9tn\u00ed techniky anal\u00fdzy dat. Uzav\u00edr\u00e1me, \u017ee v\u00fdb\u011br techniky z\u00e1vis\u00ed na tom, zda&nbsp;chce evalu\u00e1tor zjistit velikost absolutn\u00edho nebo relativn\u00edho rizika, jeho pr\u016fb\u011bh v&nbsp;\u010dase anebo pouze p\u0159\u00edtomnost nezam\u011bstnan\u00e9ho v&nbsp;registru.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Zdroje<\/strong><\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[1]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Betcherman, G., Olivas, K., Dar, A. (2004)&nbsp;<em>Impacts of Active Labor Market Programs: New Evidence from Evaluations with Particular Attention to Developing and Transition Countries<\/em>. Washington: The World Bank [online]. [cit. 2015\u201302-20]. Dostupn\u00e9 z: http:\/\/siteresources.worldbank.org\/SOCIAL<br>PROTECTION\/Resources\/SP-Discussion-papers\/Labor-Market-DP\/0402.<br>pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[2]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Bonoli, G. (2010)&nbsp;<em>The political economy of active labour market policy<\/em>. Edinburgh: RECWOWE Publication, Dissemination and Dialogue Centre [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.sps.ed.ac.uk\/__<br>data\/assets\/pdf_file\/0010\/39268\/REC-WP_0110_Bonoli.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[3]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Blache, G. (2011)&nbsp;<em>&nbsp;Active Labour Market Policies in Denmark: A Comparative Analysis of Post-Program Effects<\/em>. [online]. [cit. 2015-02-20]. Paris: CES. Dostupn\u00e9 z:&nbsp; ftp:\/\/mse.univ-paris1.fr\/pub\/mse\/CES2011\/11071.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[4]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Borland, J., Tseng, Y., Wilkins, R. (2005)&nbsp;<em>Experimental and quasi-experimental methods of microeconomic program and policy evaluation<\/em>. Melbourne: Melbourne Institute of Applied Economic and Social Research [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/cf.fbe.unimelb.edu.au\/staff\/<br>jib\/documents\/Rog_YiP_Jeff_WPJan05.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[5]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Bring, J., Kenneth Carling, K. (2001) Attrition and misclassification of&nbsp;drop-outs in the analysis of unemployment duration. Uppsala: IFAU. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.researchgate.net\/<br>publication\/5095711_Attrition_and_misclassification_of_drop-outs_in_the<br>_analysis_of_unemployment_duration<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[6]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Bryson, A., Dorsett, R., Purdon, S. (2002)&nbsp;<em>The use of propensity score matching in the evaluation of active labour market policies<\/em>. London: Policy Studies Institute and National Centre for Social Research. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/eprints.lse.ac.uk\/4993\/1\/The_use_of_propensity_<br>score_matching_in_the_evaluation_of_active_labour_market_policies.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[7]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Calmfors, L. (1994)&nbsp;<em>Active labour market policy and unemployment &#8211; a framework for the analysis of crucial design features<\/em>. Paris: OECD. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: https:\/\/search.oecd.org\/eco\/growth\/33936463<br>.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[8]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Calmfors, L., Forslund, A., Hemstr\u00f6m, M. (2002)&nbsp;<em>Does active labour market policy work? Lessons from the Swedish experiences<\/em>. Uppsala: IFAU. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.ifau.se\/upload\/pdf\/se\/2002\/<br>wp02-04.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[9]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Card, D., Kluve, J., Weber, A. (2009)&nbsp;<em>Active Labor Market Policy Evaluations: A Meta-Analysis<\/em>. Bonn: Institute for the Study of Labor. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.nber.org\/papers\/w16173<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[10]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Card, D., Ibarr\u00e1n, P., Villa, J. (2011)&nbsp;<em>Building in an Evaluation Component for&nbsp;Active Labour Market Programs: a Practitioner\u2019s Guide.<\/em>&nbsp;Washington: Inter-American Development Bank. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/publications.iadb.org\/bitstream\/handle\/11319\/5349\/Building%20in%20an%20Evalualtion%20Component%20for%20Active%20Labor%20Market%20Programs.pdf?sequence=1<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[11]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; de Koning, J., Peers, Y. (2007)&nbsp;<em>Evaluating Active Labour Market Policies Evaluations<\/em>. Rotterdam: SEOR.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[12]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Dehejia, R. H., Wahba, S. (2002) \u201ePropensity score matching methods for&nbsp;nonexperimental causal studies\u201c[online].&nbsp;<em>The review of Economics and&nbsp;Statistics<\/em>&nbsp;84(1): 151-161. [cit. 2015-02-20]. Dostupn\u00e9 z: https:\/\/wagner.<br>nyu.edu\/files\/faculty\/publications\/matching.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[13]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Dias, M., Ichimura, H., van den Berg, G. (2008)&nbsp;<em>The Matching Method for&nbsp;Treatment Evaluation with Selective Participation and Ineligibles<\/em>. Bonn: Institute for the Study of Labor. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/ftp.iza.org\/dp3280.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[14]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Fay, R. (1996).&nbsp;<em>Enhancing the effectiveness of active labour market policies: Evidence from Programme Evaluation in OECD countries<\/em>. OECD: Paris. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.oecd-ilibrary.org \/docserver\/download\/5lgsjhvj7tjl.pdf?expires=1424684458&amp;id=id&amp;accname=guest&amp;checksum=2CC5D447CE302083BB12D1CF43C8BA6E<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[15]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Forslund, A., Fredriksson, P., Vinkstr\u00f6m, J. (2011)&nbsp;<em>What active labour market policy works in recession?&nbsp;<\/em>Uppsala: IFAU [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.ifau.se\/Upload\/pdf\/se\/2011\/wp11-02-What-active-labor-market-policy-works-in-a-recession.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[16]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Heckman, J. J. and J. A. Smith (1996) \u201eExperimental and Non-Experimental Evaluation\u201c. In: Schmid, G. O&#8217;Reilly, J., Schomann K. (eds),&nbsp;<em>International Handbook of Labour Market Policy and Evaluation<\/em>. Cheltenham: Edward Elgar.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[17]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hora et al. (2009)&nbsp;<em>Hodnocen\u00ed program\u016f aktivn\u00ed politiky zam\u011bstnanosti realizovan\u00fdch v roce 2007 se zam\u011b\u0159en\u00edm na rekvalifikace (struktura, c\u00edlenost, kr\u00e1tkodob\u00e9 a st\u0159edn\u011bdob\u00e9 efekty na opu\u0161t\u011bn\u00ed evidence)<\/em>&nbsp;Praha: V\u00daPSV [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/praha.vupsv.cz\/Fulltext\/vz_300.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[18]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hora, O., Suchanec, M. (2014). \u201eZhodnocen\u00ed program\u016f aktivn\u00ed politiky zam\u011bstnanosti realizovan\u00fdch v&nbsp;\u010cesk\u00e9 republice v&nbsp;obdob\u00ed krize\u201c. In: Sirov\u00e1tka T., Hor\u00e1kov\u00e1 M., Hor\u00e1k P. (eds),&nbsp;<em>\u010cesk\u00e1 politika zam\u011bstnanosti v&nbsp;dob\u011b krize a po krizi<\/em>. Brno\/Boskovice: Masarykova Universita\/Albert. s. 143-182.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[19]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hujer, R., Caliendo, M. (2000)&nbsp;<em>Evaluation of Labour Market Policy: Methodological Concepts and Empirical Estimates<\/em>. Bonn: Institute for the Study of&nbsp;Labor [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/ftp.iza.org\/dp236<br>.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[20]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hujer, R., Wellner, M. (2000)&nbsp;<em>The Effects of Public Sector Sponzored Training on Individual Employment Performance in East Germany<\/em>. Bonn: Institute for&nbsp;the Study of Labor [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/<br>ftp.iza.org\/dp141.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[21]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hujer, R., Caliendo, M., Radi\u0107, D. (2004) \u201eMethods and Limitations of&nbsp;Evaluation and Impact Research\u201c In: Descy, P., Tessaring, M. (eds),<br><em>The Foundations of Evaluation and Impact Research<\/em>. Luxembourg: Cedefop\/Office for Official Publications of the European Communities. s. 131-190.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[22]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Hujer, R., Thomsen, S., Zeiss, Ch. (2006) The Effects of Short-Term Training Measures on the Individual Unemployment Duration in West Germany. Mannheim: ZEW [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: ftp:\/\/<br>ftp.zew.de\/pub\/zew-docs\/dp\/dp06065.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[23]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Kluve, J. et al. (2005).&nbsp;<em>Study on the effectivness of ALMPs<\/em>. Essen: RWI [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.rwi-essen.de\/media<br>\/content\/pages\/publikationen\/rwi-projektberichte\/PB_ALMP.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[24]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; LaLonde, R. J. (1986) \u201eEvaluating the Econometric Evaluations of Training Programs with Experimental Data\u201c [online].&nbsp;<em>The American Economic Review<\/em>&nbsp;76(4): 604-620. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.jstor.<br>org\/stable\/pdf\/1806062.pdf?acceptTC=true<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[25]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Lechner, M., Wunsch, C. (2009)&nbsp;<em>Active Labour Market Policy in East Germany: Waiting for the Economy to Take Off<\/em>. St. Gallen: University of St. Gallen. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www1.vwa.unisg.ch\/<br>RePEc\/usg\/dp2006\/DP24_Le.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[26]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Mohr, L. (1992)&nbsp;<em>Impact analysis for program Evaluation<\/em>. Newbury park, London, New Delphi: Sage publications.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[27]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Puhani, P. (1998)&nbsp;<em>Advantage through training? A microeconometric evaluation of the employment effects of active labour market programmes in Poland<\/em>. Leibniz: ZEW [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/papers.ssrn.com\/<br>sol3\/papers.cfm?abstract_id=141462<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[28]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Reinowski, E., Schultz, B. (2006).&nbsp;<em>Microeconometric Evaluation of Selected ESF-funded ALMP-Programmes<\/em>. Halle: Halle Institute for Economic Research [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.iwh-halle.de<br>\/d\/publik\/disc\/17-06.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[29]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Richardson, K., van den Berg, G. (2006)&nbsp;<em>Swedish Labor Market Training and&nbsp;the Duration of Unemployment<\/em>. Bonn: Institute for the Study of Labor [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/ftp.iza.org\/dp2314.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[30]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Sianesi, Barbara (2003) An evaluation of the Swedish system of Active Labour Market Programmesin the 1990s. London: The Institute for Fiscal Studies.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[31]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Smith, J., Todd P. (2005) \u201eDoes matching overcome LaLonde\u2019s critique of&nbsp;nonexperimental estimators?\u201c [online].&nbsp;<em>Journal of Econometrics<\/em>. 125(1-2): 305\u2013353. [cit. 2015-02\u201320]. Dostupn\u00e9 z: http:\/\/ac.els-cdn.com\/S030440<br>760400082X\/1-s2.0-S030440760400082X-main.pdf?_tid=e557ccd2-bb59-11e<br>4-8b56-00000aab0f02&amp;acdnat=1424695730_7b4c09dad70db27e20bd76d106<br>ccbdbf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[32]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Soukup, T. (2006)&nbsp;<em>Early assessment a profiling ve slu\u017eb\u00e1ch zam\u011bstnanosti. Zahrani\u010dn\u00ed zku\u0161enosti a spolehlivost odhadu v \u010cR<\/em>. Praha: V\u00daPSV [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/praha.vupsv.cz\/Fulltext\/vz_202.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[33]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Soukup, T., Michali\u010dka, L., Kot\u00edkov\u00e1, J. (2009)&nbsp;<em>T\u0159\u00edd\u011bn\u00ed uchaze\u010d\u016f na \u00fa\u0159adech pr\u00e1ce \u2013 \u0159e\u0161en\u00ed problematiky c\u00edlen\u00ed APZ a poradenstv\u00ed<\/em>. Praha: V\u00daPSV [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/praha.vupsv.cz\/Fulltext\/vz_287.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[34]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Suchanec, M. (2011) \u201eVyu\u017eit\u00ed Booleho p\u0159\u00edstupu v komparativn\u00ed anal\u00fdze: p\u0159\u00edklad srovn\u00e1n\u00ed 27 zem\u00ed EU z hlediska zam\u011bstnanosti \u017een\u201c.&nbsp;<em>F\u00f3rum soci\u00e1ln\u00ed politiky<\/em>5(4):17-19<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[35]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Suchanec, M. (2014a)&nbsp;<em>Jak se metodologick\u00e9 volby v&nbsp;procesu evaluace dopadu projevuj\u00ed p\u0159i evaluaci dopad\u016f \u010desk\u00e9 APZ prost\u0159ednictv\u00edm anal\u00fdzy datov\u00e9 matice OK pr\u00e1ce?<\/em>&nbsp;Brno: FSS MU.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[36]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Suchanec, M. (2014b) \u201eEvaluace dopadu\u201c. In: Hora, O., Suchanec, M., \u017d\u00ed\u017elavsk\u00fd, M.&nbsp;<em>Evalua\u010dn\u00ed v\u00fdzkum<\/em>. Brno: MUNI. s. 85-108.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[37]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; van Ours, J. (2002)&nbsp;<em>The Locking-in Effect of Subsidized Jobs<\/em>. Michigan: The&nbsp;William Davidson Institute. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/wdi.umich.edu\/files\/publications\/workingpapers\/wp474.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[38]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Vermunt, J. (1996)&nbsp;<em>Log-linear event history analysis: a general approach with&nbsp;missing data, latent variables, and unobserved heterogenity<\/em>. Tilburg: Tilburg University Press [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/members.<br>home.nl\/jeroenvermunt\/thesis.pdf<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">[39]&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Wimer, Ch. (2006)&nbsp;<em>Learning from Small-Scale Experimental Evalutions of After School Programs<\/em>. Cambridge: Harvard Graduate School of Education. [online]. [cit. 2015-02-20]. Dostupn\u00e9 z: http:\/\/www.hfrp.org\/publications-resources\/browse-our-publications\/learning-from-small-scale-experimen<br>tal-evaluations-of-after-school-programs<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref1\">[1]<\/a>&nbsp;Jedn\u00e1 se o postup hodnot\u00edc\u00ed n\u00e1sledn\u00e9 v\u00fdsledky a dopady jednotliv\u00fdch program\u016f a nikoli o procesn\u00ed evaluaci program\u016f.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref2\">[2]<\/a>&nbsp;Card, Kluve a Weber (2009) nap\u0159. v&nbsp;meta-anal\u00fdze nezjistili mezi v\u00fdsledky experiment\u00e1ln\u00edch a kvazi-experiment\u00e1ln\u00edch studi\u00ed v\u00fdznamn\u00e9 rozd\u00edly. Pro srovn\u00e1n\u00ed zji\u0161t\u011bn\u00ed rozd\u00edl\u016f v&nbsp;odhadu dopadu mezi experiment\u00e1ln\u00edm a kvazi-experiment\u00e1ln\u00edm p\u0159\u00edstupem na shodn\u00fdch individu\u00e1ln\u00edch datech srovnej LaLonde (1986), Dehejia a Wahba (2002) a Smith a&nbsp;Todd (2005). O velikosti rozd\u00edl\u016f z\u0159ejm\u011b rozhoduje p\u0159edev\u0161\u00edm kvalita p\u00e1rov\u00e1n\u00ed p\u0159\u00edpad\u016f (viz n\u00ed\u017ee).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref3\">[3]<\/a>&nbsp;Pro pot\u0159eby jednotliv\u00fdch statistick\u00fdch anal\u00fdz jsou z&nbsp;mnohorozm\u011brn\u00e9 tabulkov\u00e9 datab\u00e1ze dat OKpr\u00e1ce generov\u00e1ny anonymizovan\u00e9 v\u00fdtahy u\u017e\u0161\u00ed skupiny dat do dvourozm\u011brn\u00e9ho form\u00e1tu (datov\u00e9 matice), kde ka\u017ed\u00fd uchaze\u010d o zam\u011bstn\u00e1n\u00ed je veden na jednom \u0159\u00e1dku, a&nbsp;jednotliv\u00e9 \u00fadaje o uchaze\u010d\u00edch jsou vedeny ve sloupc\u00edch. Velikost datov\u00e9ho souboru m\u016f\u017ee b\u00fdt podle pot\u0159eb konkr\u00e9tn\u00ed anal\u00fdzy od n\u011bkolika stovek \u010di tis\u00edc\u016f p\u0159\u00edpad\u016f a\u017e po des\u00edtky tis\u00edc \u00fa\u010dastn\u00edk\u016f APZ a stovky tis\u00edc ostatn\u00edch uchaze\u010d\u016f o zam\u011bstn\u00e1n\u00ed (pro pot\u0159eby vytvo\u0159en\u00ed kontroln\u00ed skupiny) v&nbsp;jednom konkr\u00e9tn\u00edm roce. Z&nbsp;datab\u00e1ze jsou vyu\u017e\u00edv\u00e1ny p\u0159edev\u0161\u00edm informace o jednotliv\u00fdch uchaze\u010d\u00edch v\u010detn\u011b informac\u00ed o jejich historii nezam\u011bstnanosti a d\u00e1le informace o jednotliv\u00fdch programech aktivn\u00ed politiky zam\u011bstnanosti.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref4\">[4]<\/a>&nbsp;Hlavn\u00ed devizou tohoto nestatistick\u00e9ho p\u0159\u00edstupu je mo\u017enost identifikace nutn\u00fdch a dostate\u010dn\u00fdch podm\u00ednek ke vzniku po\u017eadovan\u00e9 ud\u00e1losti, a t\u00edm i identifikace potenci\u00e1ln\u00edch kl\u00ed\u010dov\u00fdch sou\u010d\u00e1st\u00ed program\u016f nutn\u00fdch k&nbsp;dosa\u017een\u00ed pozitivn\u00edho dopadu p\u0159i relativn\u011b velmi mal\u00e9 velikosti zkouman\u00e9ho vzorku. V&nbsp;p\u0159\u00edpad\u011b vy\u010derp\u00e1vaj\u00edc\u00edho \u0161et\u0159en\u00ed a velk\u00e9 velikosti souboru v\u0161ak p\u0159\u00edstup nen\u00ed obecn\u011b vhodn\u00fd, nebo\u0165 se \u00fam\u011brn\u011b velikosti vzorku zvy\u0161uje pravd\u011bpodobnost zastoupen\u00ed v\u0161ech potenci\u00e1ln\u00edch kombinac\u00ed nez\u00e1visl\u00fdch a z\u00e1visl\u00e9ho faktoru, a anal\u00fdzu tak nen\u00ed mo\u017en\u00e9 prov\u00e9st z&nbsp;d\u016fvodu ambivalence p\u016fsoben\u00ed jednotliv\u00fdch faktor\u016f. V&nbsp;p\u0159\u00edpad\u011b OKpr\u00e1ce tedy lze konfigura\u010dn\u00ed p\u0159\u00edstup doporu\u010dit: a) pokud je na\u0161\u00edm c\u00edlem identifikace kl\u00ed\u010dov\u00fdch sou\u010d\u00e1st\u00ed lok\u00e1ln\u00edch program\u016f mal\u00e9ho rozsahu u specifick\u00fdch populac\u00ed v&nbsp;konkr\u00e9tn\u00edch podm\u00ednk\u00e1ch, b) po agregaci a \u00faprav\u011b p\u016fvodn\u00edho datov\u00e9ho souboru, kdy jednotkou anal\u00fdzy nebudou nezam\u011bstnan\u00ed, ale programy a na\u0161\u00edm c\u00edlem bude nap\u0159. identifikace kl\u00ed\u010dov\u00fdch element\u016f program\u016f a vn\u011bj\u0161\u00edch podm\u00ednek skrze srovn\u00e1n\u00ed celkov\u00fdch dopad\u016f jednotliv\u00fdch programov\u00fdch cykl\u016f.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref5\">[5]<\/a>&nbsp;Jedn\u00edm ze z\u00e1kladn\u00edch evalua\u010dn\u00edch dilemat je v\u00fdb\u011br hodnocen\u00fdch \u00fa\u010dastn\u00edk\u016f programu. \u010castou situac\u00ed je zna\u010dn\u00e1 redukce dostupn\u00fdch dat (o\u0159ez\u00e1n\u00ed vzorku tak, aby splnil metodologick\u00e1 krit\u00e9ria). Evalu\u00e1tor tak z\u00edsk\u00e1 nen\u00e1hodn\u011b vybran\u00fd sub-vzorek populace, pro kter\u00fd jsou napln\u011bny jeho evalua\u010dn\u00ed p\u0159edpoklady, ale za cenu pochybnosti o nulov\u00e9m vlivu takov\u00e9 selekce respektive za cenu sn\u00ed\u017een\u00ed vn\u011bj\u0161\u00ed validity v\u00fdsledk\u016f (viz tak\u00e9 Bryson, Dorsett a Purdon 2002).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref6\">[6]<\/a>&nbsp;Jakkoli je nejvy\u0161\u0161\u00ed d\u016fraz kladen na zhodnocen\u00ed dopadu program\u016f, nelze p\u0159i politick\u00e9m rozhodov\u00e1n\u00ed ani p\u0159i interpretaci dopadu opom\u00edjet meritorn\u00ed v\u00fdznam dosa\u017een\u00e9ho v\u00fdsledku \u2013 nap\u0159. kolik nezam\u011bstnan\u00fdch si po programu nalezlo zam\u011bstn\u00e1n\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref7\">[7]<\/a>&nbsp;Dal\u0161\u00ed potencion\u00e1ln\u011b zaj\u00edmavou mo\u017enost\u00ed pro n\u011bkter\u00e9 typy v\u00fdzkumn\u00fdch anal\u00fdz je sledov\u00e1n\u00ed distribuce v\u00fdsledk\u016f u nezam\u011bstnan\u00fdch (distribu\u010dn\u00ed efekty) \u2013 viz nap\u0159. Heckman a&nbsp;Smith 1996, Hujer, Caliendo a Radi\u0107 2004. Toto rozli\u0161en\u00ed pom\u00e1h\u00e1 mj. odpov\u011bd\u011bt na&nbsp;ot\u00e1zku, zda mezi nezam\u011bstnan\u00fdmi neexistuj\u00ed v\u00fdznamn\u00e9 rozd\u00edly v dopadu z&nbsp;hlediska sledovan\u00fdch charakteristik (n\u011bkomu mohl program pomoci, zat\u00edmco jin\u00e9mu u\u0161kodit). Zahrani\u010dn\u00ed zku\u0161enosti z&nbsp;\u0159ady studi\u00ed nap\u0159. ukazuj\u00ed, \u017ee dopady program\u016f APZ se odli\u0161uj\u00ed u&nbsp;mlad\u00fdch nezam\u011bstnan\u00fdch do 25 let a dal\u0161\u00edch v\u011bkov\u00fdch skupin nezam\u011bstnan\u00fdch.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref8\">[8]<\/a>&nbsp;V&nbsp;p\u0159echoz\u00edch studi\u00edch v&nbsp;\u010cR m\u011bl v\u00fdsledek v\u017edy dichotomizovanou podobu (\u201eje v&nbsp;evidenci\u201c\/\u201enen\u00ed v&nbsp;evidenci\u201c). V\u00fdhodou tohoto p\u0159\u00edstupu je p\u0159esn\u00e9 vymezen\u00ed obou stav\u016f, ov\u0161em p\u0159i zna\u010dn\u00e9 nejistot\u011b ohledn\u011b podstaty situace \u201enen\u00ed v&nbsp;evidenci\u201c. V&nbsp;zahrani\u010d\u00ed, kde se zpravidla pracuje s&nbsp;konkr\u00e9tn\u00ed znalost\u00ed v\u00fdsledk\u016f (jako je zam\u011bstn\u00e1n\u00ed \u010di odchod na&nbsp;rodi\u010dovskou dovolenou) n\u011bkdy p\u0159etrv\u00e1v\u00e1 probl\u00e9m s&nbsp;vysok\u00fdm pod\u00edlem chyb\u011bj\u00edc\u00edch \u00fadaj\u016f (tento probl\u00e9m existuje i v&nbsp;\u010cR). Z&nbsp;tohoto hlediska je v\u00fdznamn\u00e1 diskuze, jak&nbsp;s&nbsp;t\u011bmito \u00fadaji pracovat metodologicky korektn\u00edm zp\u016fsobem, nap\u0159. kter\u00e9 stavy (situace) lze ch\u00e1pat jako trv\u00e1n\u00ed cenzorovan\u00e1 zprava (viz nap\u0159. Bring a Carling 2001, Richardson a van den Berg 2006).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref9\">[9]<\/a>&nbsp;Nap\u0159. pot\u0159ebnost nezam\u011bstnan\u00fdch, vzd\u011bl\u00e1vac\u00ed efekt programu, (ne)vhodnost hodnocen\u00ed efektu, kter\u00fd nemohl dosud nastat.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref10\">[10]<\/a>&nbsp;Respektive ve srovn\u00e1n\u00ed se situac\u00ed ne\u00fa\u010dasti v&nbsp;programu.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref11\">[11]<\/a>&nbsp;K&nbsp;efektu sl\u00edz\u00e1v\u00e1n\u00ed smetany doch\u00e1z\u00ed, pokud jsou do programu vybr\u00e1ni \u00fa\u010dastn\u00edci, kte\u0159\u00ed by nejpravd\u011bpodobn\u011bji usp\u011bli i bez programu na \u00fakor nezam\u011bstnan\u00fdch, kter\u00fdm by program mohl nejv\u00edce pomoci (viz nap\u0159. Blache 2011).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref12\">[12]<\/a>&nbsp;Jednotliv\u00e1 za\u0159azen\u00ed do programu se tedy pravd\u011bpodobn\u011bji budou odli\u0161ovat, pokud osoby do programu za\u0159azuj\u00ed r\u016fzn\u00e9 osoby, a ka\u017ed\u00e1 z&nbsp;nich vyu\u017e\u00edv\u00e1 jin\u00e1 krit\u00e9ria pro za\u0159azen\u00ed nebo tato krit\u00e9ria jinak interpretuje, anebo pokud sami nezam\u011bstnan\u00ed mohou rozhodovat, zda se programu z\u00fa\u010dastn\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref13\">[13]<\/a>&nbsp;I v&nbsp;designu prav\u00e9ho experimentu m\u016f\u017ee nastat selekce z&nbsp;d\u016fvodu randomizace a\/nebo&nbsp;substituce (Suchanec 2014b).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref14\">[14]<\/a>&nbsp;Jedn\u00edm z alternativn\u00edch postup\u016f pro \u0159e\u0161en\u00ed probl\u00e9mu selekce m\u016f\u017ee b\u00fdt nespojit\u00e1 regrese (regression discontinuity design). Jej\u00ed vyu\u017eit\u00ed vy\u017eaduje ov\u0161em sp\u00ed\u0161e specifick\u00e9 podm\u00ednky, kter\u00e9 je obt\u00ed\u017en\u00e9 naplnit. Pro vysv\u011btlen\u00ed zp\u016fsobu, jak se nespojit\u00e1 regrese vypo\u0159\u00e1d\u00e1v\u00e1 s&nbsp;probl\u00e9mem selekce, viz Suchanec (2014b), pro konkr\u00e9tn\u00ed vyu\u017eit\u00ed p\u0159i evaluaci dopadu viz nap\u0159. Dias, Ichimura a van den Berg (2008).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref15\">[15]<\/a>&nbsp;Z\u00e1kladn\u00edmi zp\u016fsoby jsou p\u0159esn\u00e9 p\u00e1rov\u00e1n\u00ed (exact matching), p\u00e1rov\u00e1n\u00ed \u201enejbli\u017e\u0161\u00edho souseda\u201c (nearest neighbour matching), p\u00e1rov\u00e1n\u00ed nejbli\u017e\u0161\u00edho souseda ve vymezen\u00e9m intervalu (caliper matching) \u010di v\u00e1\u017een\u00ed p\u0159\u00edpad\u016f na z\u00e1klad\u011b jejich podobnosti s&nbsp;p\u00e1rovan\u00fdm p\u0159\u00edpadem (kernel matching) \u2013 viz nap\u0159. Hujer a Wellner 2000, Bryson, Dorsett a Purdon 2002, Hujer, Caliendo a Radi\u0107 2004. V&nbsp;p\u0159\u00edpad\u011b OKpr\u00e1ce dosud v\u017edy bylo mo\u017en\u00e9 p\u00e1rovat na p\u0159esn\u00e9m propensity score, co\u017e je ve srovn\u00e1n\u00ed s&nbsp;p\u00e1rov\u00e1n\u00edm v&nbsp;intervalu p\u0159\u00edstup,<br>kter\u00fd by m\u011bl nejv\u00edce redukovat potencion\u00e1ln\u00ed zkreslen\u00ed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref16\">[16]<\/a>&nbsp;V&nbsp;praxi jsou pro p\u00e1rov\u00e1n\u00ed v\u00fdznamn\u00e1 dv\u011b z\u00e1kladn\u00ed hlediska: a) velikost skupiny potenci\u00e1ln\u00edch partner\u016f pro p\u00e1rov\u00e1n\u00ed do kontroln\u00ed skupiny, b) definovan\u00e1 krit\u00e9ria pro pot\u0159ebnou bl\u00edzkost hodnot propensity score v&nbsp;r\u00e1mci jednoho p\u00e1ru. V&nbsp;na\u0161em p\u0159\u00edpad\u011b je mno\u017estv\u00ed p\u0159\u00edpad\u016f vyu\u017eiteln\u00fdch pro p\u00e1rov\u00e1n\u00ed velmi vysok\u00e9, a proto nemus\u00edme p\u0159\u00edpady do p\u00e1rov\u00e1n\u00ed za\u0159azovat opakovan\u011b a p\u0159esto z\u00edsk\u00e1v\u00e1me v\u011bt\u0161inu p\u00e1r\u016f s&nbsp;p\u0159esnou shodou na propensity score.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref17\">[17]<\/a>&nbsp;V&nbsp;p\u0159edchoz\u00ed anal\u00fdze (Hora a Suchanec 2014) jsme p\u0159i pom\u011brn\u011b komplexn\u00edm zp\u016fsobu p\u00e1rov\u00e1n\u00ed nesp\u00e1rovali asi 15 procent p\u0159\u00edpad\u016f. Dodate\u010dn\u00e1 anal\u00fdza uk\u00e1zala, \u017ee tyto p\u0159\u00edpady dosahovaly hor\u0161\u00edch pr\u016fm\u011brn\u00fdch m\u00edstn\u00edch v\u00fdsledk\u016f ne\u017e 85 procent sp\u00e1rovan\u00fdch p\u0159\u00edpad\u016f. U t\u011bchto p\u0159\u00edpad\u016f zpravidla nen\u00ed mo\u017en\u00e9 naj\u00edt srovnateln\u00fd p\u0159\u00edpad v&nbsp;kontroln\u00ed skupin\u011b. Alternativn\u00edm p\u0159\u00edstupem by proto mohlo b\u00fdt: a) up\u0159ednostnit tyto p\u0159\u00edpady p\u0159i p\u00e1rov\u00e1n\u00ed, b) uvolnit u nich krit\u00e9rium p\u0159esnosti odhadu v&nbsp;propensity score, c) povolit v\u00edcen\u00e1sobn\u00e9 pou\u017eit\u00ed p\u0159\u00edpadu. Z\u00e1kladn\u00ed evalua\u010dn\u00ed ot\u00e1zkou je, zda chceme p\u0159esn\u00fd odhad dopadu o 85 procentech p\u0159\u00edpad\u016f \u010di m\u00e9n\u011b p\u0159esn\u00fd odhad u v\u011bt\u0161\u00edho pod\u00edlu p\u0159\u00edpad\u016f.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref18\">[18]<\/a>&nbsp;N\u011bkte\u0159\u00ed auto\u0159i interpretuj\u00ed tyto informace v&nbsp;jin\u00e9m kontextu, a to jako na\u010dasov\u00e1n\u00ed intervence (viz nap\u0159. Hujer, Thomsen a Zeiss 2006).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref19\">[19]<\/a>&nbsp;P\u0159edchoz\u00ed kvazi-experiment\u00e1ln\u00ed studie uk\u00e1zaly, \u017ee kr\u00e1tkodob\u00e9 a dlouhodob\u00e9 dopady program\u016f APZ se mohou odli\u0161ovat (kr\u00e1tkodob\u00e9 dopady byly m\u00e9n\u011b p\u0159\u00edzniv\u00e9 ne\u017e dlouhodob\u00e9 dopady) (viz nap\u0159. Calmfors, Forslund a Hemstr\u00f6m 2002, Lechner a Wunsch 2009, Card, Kluve a Weber 2009). Opa\u010dn\u00e1 situace m\u016f\u017ee nastat u program\u016f tvorby m\u00edst (viz Reinowski a Schultz 2006). Z&nbsp;tohoto d\u016fvodu je doporu\u010dov\u00e1no m\u011b\u0159it v\u00fdsledky ve v\u00edce \u010dasov\u00fdch bodech v\u010detn\u011b bod\u016f minim\u00e1ln\u011b dva roky po absolvov\u00e1n\u00ed programu (nap\u0159.&nbsp;Card, Kluve a Weber 2011).<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/www.evaltep.cz\/inpage\/evaluace-dopadu-apz\/#_ftnref20\">[20]<\/a>&nbsp;Podle Vermunta (1996) pravd\u011bpodobnost p\u0159e\u017eit\u00ed (\u010di funkce p\u0159e\u017eit\u00ed) indikuje pravd\u011bpodobnost, \u017ee jev nenastane v&nbsp;\u010dase t. M\u00edra hazardu (\u010di funkce hazardu) vyjad\u0159uje nyn\u011bj\u0161\u00ed riziko, \u017ee ud\u00e1lost nastane v&nbsp;\u010dase t, pokud ji\u017e nenastala v&nbsp;d\u0159\u00edv\u011bj\u0161\u00edm obdob\u00ed. M\u00edra hazardu m\u016f\u017ee dosahovat hodnot vy\u0161\u0161\u00edch ne\u017e jedna. Obdobn\u00e1 m\u00edra je vyu\u017e\u00edv\u00e1na i v&nbsp;modelech s&nbsp;nespojit\u00fdm \u010dasem pod n\u00e1zvem podm\u00edn\u011bn\u00e1 pravd\u011bpodobnost (ibid.).<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Realization of impact evaluation brings possibility of various methodological choices that are limited to distinct assumptions concerning the context of the study, nature of the data etc. Results of the impact evaluation are then dependent considerably on definition of particular evaluation model. Aim of this article is to discuss main areas, principles and approaches to evaluation of Czech active labor policy programs (ALMP) outcomes and impacts, using administrative data of OKpr\u00e1ce. The article presents approaches that are suitable for this kind of data and is based on partial equilibrium perspective, thus omitting impacts on labor market as whole. The core of the article is discussion of diverse methodical steps in context of internal validity of measured impact. The matching is discussed as a method of balancing pre-program differences between program participants and control group. Inclusion of factors of labor market perspective and history, as well as phase of economic cycle, during this process is of high importance. Article concludes that choice of particular statistical method of data analysis hinges on whether evaluator wants to find absolute or relative risks of leaving unemployment, their progress in time or merely the presence of unemployed in register.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[26],"tags":[],"class_list":["post-1403","post","type-post","status-publish","format-standard","hentry","category-articles"],"acf":[],"_links":{"self":[{"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/posts\/1403","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/comments?post=1403"}],"version-history":[{"count":2,"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/posts\/1403\/revisions"}],"predecessor-version":[{"id":1405,"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/posts\/1403\/revisions\/1405"}],"wp:attachment":[{"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/media?parent=1403"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/categories?post=1403"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/evaltep.xcreative.cz\/en\/wp-json\/wp\/v2\/tags?post=1403"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}