Table of Contents


Attrition Profiles

A total of 4,506 (71.8% HFI and 28.8% comparison) families were referred from 40 HFI programs throughout the state.  This volume of participation serves as an indicator of the impressive research-practitioner collaboration that was achieved over the course of this project.  Further, the widespread involvement of HFI programs serves as an indicator of the degree to which HFI programs are committed to demonstrating that their programs have a significant influence in the lives of the children and families they serve.

Attrition at Time One.

Of course in a highly mobile, hard to reach population, not all families referred were successfully engaged in the data collection process.  Of the 4,506 familes referred, 280 (6.2%) were still pending data collection at the close of the evaluation project.  Further 125 (2.8%) families were deemed inappropriate referrals (e.g., children too old to be included in Time 1 data collection at the time of referral or consent from a legal guardian of a minor participant was not obtained).

Of the remaining 4,101 viable referrals Time 1 data was collected from 2,220 (54.1%).  This rate of engagement is comparable to that obtained in other large scale Healthy Families evaluations.

The reasons for not obtaining Time 1 data from the remaining referrals are described in the figure below.  The most common reason for not obtaining Time 1 data resulted when interviewers were unable to locate the family referred.  In these cases interviewers made numerous attempts to contact the families (multiple phone calls, inquiries with referring agencies, letters) but were unable to reach the parents to arrange for the first data collection appointment.

The second most common reason for not obtaining Time 1 data was that the family declined participation when contacted to set up an appointment.  This withdrawl rate, although small (i.e., 5.2%), helps to verify that parents understood that their participation was voluntary and that they could withdraw at any time without penalty or prejudice.

Of the 4,101 viable Time 1 referrals, 6.1% aged out (i.e. were older than 3 months of age) before an interviewer was able to collect Time 1 data.  Inability to reach a viable referral on time resulted for a variety of reason including late referrals (i.e., receipt of the referral with less than one month to collect data), temporary lack of coverage in a service area, or interviewers failed to schedule data collection in a timely fashion.

Figure 1: Attrition at Time One

Status Percent
Collected 54.1%
Couldn't Locate ~ 21.9%
Declined 5.2%
Moved 4.8%
Aged Out 6.1%
Dropped Intervention 6.1%
Other 1.8%

Other reasons for not obtaining Time 1 data included the fact that a number of familes had already moved out of the HFI service area (4.8%) or had dropped out of the HFI program (6.1%) before their Time 1 data were due to be collected.

A small number of cases (1.8%) were dropped for "other" reasons including adverse pregnancy outcomes (miscarriages, stillbirths, etc).

Attrition across Two Years

The attrition patterns for Times 2, 3, 4, and 5 are presented in the table below.  Because the goal of the data collection process was to obtain as many families as possible with data at all five timepoints, only cases for which the previous time point had been collected were considered viable cases at subsequent time points.  For example, only cases in which time 1 data were obtained were assigned for data collection at Time 2.  This allowed project resources to be focused on those cases for which complete data would be available for the entire two year period.

As may be seen in the table below, the majority (approximately 61.9%) of viable cases were collected at each time point.  On average, data were not obtained from 7.4% of the viable cases at each time point because the project ended before the data were due to be collected.  As was the case at Time 1, the most common reason for not collecting the follow-up data was inability to locate participants despite repeated efforts to contact them.  The next most common reason for not obtaining data at follow-up time points was that participants were no longer receiving HFI services.  To determine whether parents were continunig to receive services, referring agencies were contacted monthly and asked whether participating parents were still engaged in the HFI program.  Parents who were on creative outreach were assigned for continued data collection; however parents whose cases were closed were discontinued.

On average 3.5% of families were dropped at each follow up time point because the families had moved and were no longer receiving HFI services. For approximately 3.9% of families at each followup time point data were not obtained because the in-home interviews could not be completed within the appropriate time frame (i.e., they aged out).  In approximately 1.7% of the cases at each time point data were not obtained for some other reason (e.g., mother died, baby adopted).

Status 6 Months 1 year 18 months 2 years
Collected 59.0 60.0 62.3 66.2
Evaluation Project Ended 6.5 5.4 8.8 9.0
Couldn't Locate 12.2 11.1 9.3 7.2
Declined 2.1 2.0 0.9 0.6
Moved 5.1 4.3 2.4 2.2
Aged Out 4.7 4.0 3.6 3.5
Dropped intervention 9.6 11.8 10.9 8.4
Other Reasons 0.8 1.4 1.8 2.9

As is the case in any longitudinal evaluation, generalizability of findings may be limited to the extent that there were differences in the characteristics of the families that completed, versus those that did not complete, the two-year study.

To examine this issue, a series of correlational analyses were conducted to examine the associations between number of data collection sessions completed and the demographic variables listed in the table to the right. Positive correlations indicate that a given factor was associated with remaining in the study longer.  All correlations were small (all r's less than .07). Although two variables, mother's age and child's sex, were statistically significant in these large samples, neither variables accounted for more than one percent of the variance in explaining how many sessions were attended.  Thus, none of the demographic factors examined were considered to be meaningfully associated with attrition from the sample.  Based on these analyses, there is no reason to believe that families that completed the two year project were demographically different from those that did not complete the study.

Demographic Characteristic N Correlation
Mother's age 2,036 0.07*
Father's age 1,941 0.06
Child's age 2,038 0.03
Mother's education 2,035 0.04
Father's education 1,787 0.04
Child's sex 2,038 0.06*
Child first born 2,338 -0.04
Mother student 2,029 -0.01
Father student 1,878 -0.01
Mother employed 2,014 0.02
Father employed 1,818 0.04
Household size 2,038 -0.04
Chicago vs. downstate 2,038 0.01

* p less than .01