Áhrif útsendingartíma á svörun í netkönnun

Abstract

Með lækkandi svarhlutfalli í netkönnunum er mikilvægt að greina hvernig hægt sé að hámarka svarhlutfall. Í þessari rannsókn var kannað hvort ólíkir útsendingardagar og útsendingartímar könnunar hefðu áhrif á svarhlutfall meðlima í netpanel, auk þess sem áhrif bakgrunnsþátta voru greind. Notast var við gögn frá 10.446 meðlimum netpanels, eða alls 57.447 þátttökuboða, úr ferilkönnun þar sem könnun var send á 200-1000 manna slembivalið úrtak daglega í nærri heilt ár. Áhrifin voru greind með aðhvarfsgreiningu og sýndu niðurstöður að vikudagur útsendingar hefur einhver, þó lítil, áhrif á heildarsvarhlutfall. Áhrifin voru hins vegar talsverð á svörun innan 24ra og 62ja klukkustunda frá þátttökuboði. Allra síst ætti að senda könnun út á föstudegi og helst á sunnudögum eða mánudögum. Tími útsendingar hafði þau áhrif að örlítið dregur úr svörun ef þátttökuboð berst á kvöldin, þá sérstaklega undir lok vikunnar, auk þess sem svartími lengist ef sent er út á kvöldin. Ekki var munur á því hvenær fólk svaraði eftir bakgrunnsþáttum, nema að þeir sem eru með vinnu-/nemendanetfang svara síður á föstudögum og um helgar. Frekari rannsókna er þörf á áhrifum annarra bakgrunnsþátta á svörun ólíkra daga auk þess sem kanna þarf tímasetningu útsendinga betur.

With continuing lowering response rates in web surveys it is important to identify what factors can increase response rates. In this thesis the effects of sending surveys on different weekdays and time of day on response rates of members of an online panel were studied, as well as studying the effect of demographic factors. The data consisted of information on 10.466 panel members, totalling in 57.447 invitations to a web-based tracking survey which spanned for almost a year by sending daily surveys to random samples of 200-1000 members. Regression analysis was used for analysis and showed that there were some, though small, effects of weekdays on response rates. The effects were considerable when considering response rates within 24 or 62 hours from sending the invitation. Friday is the least preferred day to send out a survey with Sunday and Monday being preferred. The time of day of sending invitations had the effects that sending during evenings might decrease response rates somewhat, especially when sent at the end of the week, as well as creating longer response times. Demographic factors had no effect on what day people answered surveys, but those who had their working email registered in the panel were less likely to answer on weekends. More research is needed on the effects of further demographic factors as well as the effects of time of day need to be researched further.