![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
rivkaI just sent my dissertation advisor an e-mail describing the results of my first pass through the data analysis and asking where he thinks I should go from here. It occurred to me that maybe some of my booster section - all of whom are too kind or too well-trained to actually ask me how my dissertation's going - would like to know about the results as they emerge.
Very quick summary/review: My dissertation is about whether children with physical disabilities are at greater risk of parental abuse, and which parent, child, and family factors may affect the likelihood that parents will abuse their disabled children. My major outcome variable is the likelihood that parents will select physical punishment responses to slides of children misbehaving. (See explanation here)
Hi John,
Currently I've done some analyses on the first 65 subjects.(Martha's [my undergrad research assistant, and the only good one I've had so far - out of three I've trained] run 2 more whose data I haven't yet received, and she's trying to schedule 3 additional subjects.) [All my data collection needs to be done by April 14, because that's when federal data privacy laws change and I'd have to redo all my forms. I'd like to have a total of 70 subjects when I'm done.] Here's a quick narrative (mostly number-free) account of what I've found so far:
* The two groups [parents of at least one disabled child vs. parents of healthy children] are comparable on most demographic variables: sex, race, economic group, child's age, maternal age at child's birth, family structure (single parent, stepfamily, etc.), and number of siblings.
[It's critically important to ensure that your variable of interest (in my case, child disability) is the only significant difference between the groups. A lot of studies on this topic have failed to do so - for example, several studies have shown that premature infants are at greater risk of abuse later on, but they didn't control for the fact that preemies are more likely to be born to teenage mothers, low-income parents, and mothers who use drugs and/or alcohol, all of whom are more likely to abuse their children regardless of prematurity. At any rate, it's a Good Thing that my groups are very similar on these background factors.]
There is a significant difference in geographical region - parents of disabled children were more likely to describe themselves as living in a rural area. That's not surprising given the nature of the recruitment sites, but it is inconvenient.
[Parents of disabled children come from all over Iowa and western Illinois to have their children evaluated at the University of Iowa Health Center, which is why so many rural parents were included in my sample. My comparison group of parents of healthy children were recruited from the general pediatrics department, so it pretty much just included people who live in and around Iowa City. This is a little problem, but not a huge one, because unless you start talking about inner cities you don't see a lot of variation in abuse risk by region.]
* There are significant differences between the two groups on the Analog Parenting Task. [That's the disciplinary measure.]
[This is a huge relief: it's what I... well, not hoped to find, because it's a depressing result, but it's what I was trying to prove.]
Most importantly, when all scenes are aggregated, parents of disabled children are significantly more likely to endorse physical punishment as an initial response. No differences emerged for harsh verbal punishment (e.g., yelling, ridicule), or for escalated punishment, when all scenes were aggregated.
When different scene types are considered separately, parents of disabled children are significantly more likely to *escalate* to physical punishment in response to normative child behaviors. This is particularly striking because no parent of a healthy child endorsed physical escalation for normative behaviors. Parents of healthy children were, however, more likely than parents of disabled children to escalate to harsh verbal punishment in response to normative behaviors.
[Normative behaviors are things that are normal for kids to do, even if parents are likely to find them irritating - like a six year old girl putting on her mother's makeup, or a toddler spilling out a sewing basket. This is in contrast to child behaviors that are dangerous or destructive.]
* The two groups were also significantly different on almost everything else: [meaning the parent, child, and family factors I thought might be related to abuse risk.]
all Vineland scales, of course, ["of course" because the Vineland is how I assessed disability. The full name is the "Vineland Adaptive Behavior Scale," and it's an interview measure in which the parent is asked to describe the child's level of functioning in different areas of everyday life: dressing, for example, and talking, and interacting socially with others. There are four parts: Communication, Daily Living Skills, Socialization, and Motor Skills. Children receive scores based on how they compare to other children their age. I had a wide sampling of abilities among my disabled kids - some were within the normal range developmentally, and some could do almost nothing.]
all scales of the Questionnaire on Resources and Stress, [this was my measure of parenting stress. It includes questions about how burdened the parent feels by the child, the impact they feel the child has on family life, how much care the child needs, and how unhappy and pessimistic the parent is.]
the Attachment Q-sort correlation with the ideal sort, [okay, this is going to take some explanation. Loosely speaking, "attachment" refers to the strength of the parent-child relationship, particularly from the child's perspective. The Attachment Q-sort is a set of cards describing different aspects of child personality and behavior, all of which are seen as either evidence of attachment or facilitators of attachment. The parent sorts the cards into piles based on how well they describe their child. Then we compare the way the parent sorted the cards to a theoretically "ideal" sort. The closer the parent's sort matches the ideal sort, the better the attachment relationship.]
most of the subscales of the Attachment Q-sort, [which break child behavior and personality into specific categories - for example, whether the child enjoys physical contact with the parent, or whether the child is compliant with parental requests.]
and PEAS positive social activities. [the PEAS is an activity schedule: the parent simply marks which positive and negative activities they've done in the past week. It gives a snapshot of the parent's life. I found, for example, that the disabled group included a subgroup of parents who reported almost no positive social interactions whatever - no conversations, no hugs or kisses, no social gatherings.]
Each of these differences was in the expected direction - parents of disabled children fared worse.
* As we'd discussed, I looked for correlations between APT scores and Vineland scores. [I was trying to see whether there was a relationship between how impaired a child was, and how much they were at risk for abuse. There's reason to believe, based on other people's data, that children who are mildly disabled are at the greatest risk - except for communication disabilities (like deafness), in which case the greater the disability the greater the risk.] For the whole sample (disabled and healthy together), Vineland scores were moderately negatively correlated with physical punishment scores (for example, Vineland ABC is correlated -.341, p=.005) - but I'm not confident that this provides any evidence that degree of disability severity is related to likelihood of abuse. It may just reflect the group difference. When the disabled group alone was considered, there was no significant correlation between Vineland scores and punishment. I think this is probably an area of the data that needs more examination.
Am I correct in assuming that, without a significant relationship between disability severity and physical punishment, I can't test the mediating/moderating role of attachment and/or stress? [I want to know whether, given the presence of a disability, parents who have a more positive attachment relationship with their child and are under less stress are less likely to make abusive discipline choices.]
* Another interesting area I've been looking at is which factors correlate with a "more ideal" Attachment Q-sort within the disabled group. There are some surprising absences - for example, the Vineland Communication scale doesn't correlate - but other things do: QRS Negative Affect, QRS Child Problems (which reflects the parent's belief that the child has a negative impact on the family), Vineland Socialization, and overall child incapacity (QRS incapacity, Vineland ABC). [I think all of this makes adequate sense, at least if you know that "affect" is essentially a technical term for "mood."] That last is probably just a reflection of the fact that the most severely disabled children weren't capable of a lot of the affiliative movements and actions considered typical of the ideal sort. [Which could be theoretically important, because children's actions are very important in the formation of a positive parent-child relationship, even in infancy.]
It seems that this is a very rich data set with a lot going on. I'd really appreciate your guidance about how to proceed with the analyses so that I'm extracting as much useful information as possible. Please let me know what you want me to send you, and where you think I should go from here.
Rebecca
Very quick summary/review: My dissertation is about whether children with physical disabilities are at greater risk of parental abuse, and which parent, child, and family factors may affect the likelihood that parents will abuse their disabled children. My major outcome variable is the likelihood that parents will select physical punishment responses to slides of children misbehaving. (See explanation here)
Hi John,
Currently I've done some analyses on the first 65 subjects.(Martha's [my undergrad research assistant, and the only good one I've had so far - out of three I've trained] run 2 more whose data I haven't yet received, and she's trying to schedule 3 additional subjects.) [All my data collection needs to be done by April 14, because that's when federal data privacy laws change and I'd have to redo all my forms. I'd like to have a total of 70 subjects when I'm done.] Here's a quick narrative (mostly number-free) account of what I've found so far:
* The two groups [parents of at least one disabled child vs. parents of healthy children] are comparable on most demographic variables: sex, race, economic group, child's age, maternal age at child's birth, family structure (single parent, stepfamily, etc.), and number of siblings.
[It's critically important to ensure that your variable of interest (in my case, child disability) is the only significant difference between the groups. A lot of studies on this topic have failed to do so - for example, several studies have shown that premature infants are at greater risk of abuse later on, but they didn't control for the fact that preemies are more likely to be born to teenage mothers, low-income parents, and mothers who use drugs and/or alcohol, all of whom are more likely to abuse their children regardless of prematurity. At any rate, it's a Good Thing that my groups are very similar on these background factors.]
There is a significant difference in geographical region - parents of disabled children were more likely to describe themselves as living in a rural area. That's not surprising given the nature of the recruitment sites, but it is inconvenient.
[Parents of disabled children come from all over Iowa and western Illinois to have their children evaluated at the University of Iowa Health Center, which is why so many rural parents were included in my sample. My comparison group of parents of healthy children were recruited from the general pediatrics department, so it pretty much just included people who live in and around Iowa City. This is a little problem, but not a huge one, because unless you start talking about inner cities you don't see a lot of variation in abuse risk by region.]
* There are significant differences between the two groups on the Analog Parenting Task. [That's the disciplinary measure.]
[This is a huge relief: it's what I... well, not hoped to find, because it's a depressing result, but it's what I was trying to prove.]
Most importantly, when all scenes are aggregated, parents of disabled children are significantly more likely to endorse physical punishment as an initial response. No differences emerged for harsh verbal punishment (e.g., yelling, ridicule), or for escalated punishment, when all scenes were aggregated.
When different scene types are considered separately, parents of disabled children are significantly more likely to *escalate* to physical punishment in response to normative child behaviors. This is particularly striking because no parent of a healthy child endorsed physical escalation for normative behaviors. Parents of healthy children were, however, more likely than parents of disabled children to escalate to harsh verbal punishment in response to normative behaviors.
[Normative behaviors are things that are normal for kids to do, even if parents are likely to find them irritating - like a six year old girl putting on her mother's makeup, or a toddler spilling out a sewing basket. This is in contrast to child behaviors that are dangerous or destructive.]
* The two groups were also significantly different on almost everything else: [meaning the parent, child, and family factors I thought might be related to abuse risk.]
all Vineland scales, of course, ["of course" because the Vineland is how I assessed disability. The full name is the "Vineland Adaptive Behavior Scale," and it's an interview measure in which the parent is asked to describe the child's level of functioning in different areas of everyday life: dressing, for example, and talking, and interacting socially with others. There are four parts: Communication, Daily Living Skills, Socialization, and Motor Skills. Children receive scores based on how they compare to other children their age. I had a wide sampling of abilities among my disabled kids - some were within the normal range developmentally, and some could do almost nothing.]
all scales of the Questionnaire on Resources and Stress, [this was my measure of parenting stress. It includes questions about how burdened the parent feels by the child, the impact they feel the child has on family life, how much care the child needs, and how unhappy and pessimistic the parent is.]
the Attachment Q-sort correlation with the ideal sort, [okay, this is going to take some explanation. Loosely speaking, "attachment" refers to the strength of the parent-child relationship, particularly from the child's perspective. The Attachment Q-sort is a set of cards describing different aspects of child personality and behavior, all of which are seen as either evidence of attachment or facilitators of attachment. The parent sorts the cards into piles based on how well they describe their child. Then we compare the way the parent sorted the cards to a theoretically "ideal" sort. The closer the parent's sort matches the ideal sort, the better the attachment relationship.]
most of the subscales of the Attachment Q-sort, [which break child behavior and personality into specific categories - for example, whether the child enjoys physical contact with the parent, or whether the child is compliant with parental requests.]
and PEAS positive social activities. [the PEAS is an activity schedule: the parent simply marks which positive and negative activities they've done in the past week. It gives a snapshot of the parent's life. I found, for example, that the disabled group included a subgroup of parents who reported almost no positive social interactions whatever - no conversations, no hugs or kisses, no social gatherings.]
Each of these differences was in the expected direction - parents of disabled children fared worse.
* As we'd discussed, I looked for correlations between APT scores and Vineland scores. [I was trying to see whether there was a relationship between how impaired a child was, and how much they were at risk for abuse. There's reason to believe, based on other people's data, that children who are mildly disabled are at the greatest risk - except for communication disabilities (like deafness), in which case the greater the disability the greater the risk.] For the whole sample (disabled and healthy together), Vineland scores were moderately negatively correlated with physical punishment scores (for example, Vineland ABC is correlated -.341, p=.005) - but I'm not confident that this provides any evidence that degree of disability severity is related to likelihood of abuse. It may just reflect the group difference. When the disabled group alone was considered, there was no significant correlation between Vineland scores and punishment. I think this is probably an area of the data that needs more examination.
Am I correct in assuming that, without a significant relationship between disability severity and physical punishment, I can't test the mediating/moderating role of attachment and/or stress? [I want to know whether, given the presence of a disability, parents who have a more positive attachment relationship with their child and are under less stress are less likely to make abusive discipline choices.]
* Another interesting area I've been looking at is which factors correlate with a "more ideal" Attachment Q-sort within the disabled group. There are some surprising absences - for example, the Vineland Communication scale doesn't correlate - but other things do: QRS Negative Affect, QRS Child Problems (which reflects the parent's belief that the child has a negative impact on the family), Vineland Socialization, and overall child incapacity (QRS incapacity, Vineland ABC). [I think all of this makes adequate sense, at least if you know that "affect" is essentially a technical term for "mood."] That last is probably just a reflection of the fact that the most severely disabled children weren't capable of a lot of the affiliative movements and actions considered typical of the ideal sort. [Which could be theoretically important, because children's actions are very important in the formation of a positive parent-child relationship, even in infancy.]
It seems that this is a very rich data set with a lot going on. I'd really appreciate your guidance about how to proceed with the analyses so that I'm extracting as much useful information as possible. Please let me know what you want me to send you, and where you think I should go from here.
Rebecca
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2003-03-28 02:07 pm (UTC)no subject
Date: 2003-03-28 02:08 pm (UTC)no subject
Date: 2003-03-28 02:20 pm (UTC)But I'm fascinated whenever you share stuff like this.
no subject
Date: 2003-03-28 02:56 pm (UTC)no subject
Date: 2003-03-28 02:28 pm (UTC)no subject
Date: 2003-03-28 02:28 pm (UTC)But mine gut makes lousy dissertation material. Interesting stuff you have there.
no subject
Date: 2003-03-28 02:54 pm (UTC)And yes, generally speaking, socially isolated parents tend to report higher stress levels and to be at greater risk of taking it out on their kids. Which is not to say that the majority of them do - it's just a risk factor.
no subject
Date: 2003-03-29 02:20 pm (UTC)no subject
Date: 2003-03-28 04:31 pm (UTC)WHOOOP!!!
Date: 2003-03-28 07:59 pm (UTC)no subject
Date: 2003-03-29 06:19 am (UTC)And let me know if you ever want to commisserate about depressing dissertation results!
-J
no subject
Date: 2003-03-29 07:02 am (UTC)no subject
Date: 2003-03-29 03:31 pm (UTC)no subject
Date: 2003-03-31 06:35 am (UTC)