Need help interpreting research statistics...

ImThatGuy, BSN, RN · Apr 30, 2011

I've had statistics and research methods, and it still doesn't mean anything to me, lol.

delilas · Apr 30, 2011

Thanks for trying :) I've been attempting to read statistic guides all afternoon and simply can't make sense of it.

I had a back up paper topic, but unfortunately, it looks like the data reporting was done the same way :anbd:

Whispera, MSN, RN · Apr 30, 2011

There should be an abstract at the beginning of the article and a conclusion at the end. That should give you lots of information. I'm assuming this is in a journal that formats the research they present this way.

Your first priority in writing your report is to determine what is being measured. You didn't tell us that. I see pre-operative data...about what? Is there therapeutic play involved and the researcher measured something pre- and post-operatively?

You can do this without t and p values. They could be overkill if no one understands what they mean anyway, in your class. What is the course? Research analysis? Maternal/child nursing? Your focus would be different for each. You wouldn't necessarily be expected to be able to analyze advanced statistics if you're not in a research analysis course or if you haven't had that course or statistics.

So...what is being measured? Who was being studied? Ages? Gender? Illness? How many in each group? What was manipulated in the study? How was it being manipulated (what was the therapeutic play, if that's the focus?)? How did the scores change in the experimental group? How did those compare to the scores in the control group? Did the research show that the therapeutic play had an impact? Those are the basics... Again, this should be summarized in the article.

Is this a written report? Oral? Is the instructor's intention to help you learn more about maternal/child nursing or about statistics? That intention matters. You might want to talk to or email your instructor about all of this....

delilas · Apr 30, 2011

The measurement is the level of anxiety in children who engage in therapeutic play pre-op versus children who do not.

The abstract does help, but the conclusion is simply that anxiety levels were lower on children and adults who engaged in therapeutic play. They give the same mean and SD values in the conclusion.

It's only a maternal/child nursing class; we've really had no outright experience in statistics and research beyond simple research projects thus far, but this one requires the use of a randomized controlled study as the focus.

I wouldn't be as concerned about the numbers and interpreting them, but we're expected to create charts for the written report to show the data in a visual way.

I will email my instructor, but I know I wouldn't get any answers until late Monday evening, and really wanted to get done with it this week. I'll keep digging around CINAHL and the like for further topics, but had wanted to use this one.

nurse.sandi · Apr 30, 2011

I just finished a 7000 level Quantitative methods and analysis class. We need an alpha level. What is that set up as? Most of the time the alpha is set at .05. With a p=.001, you probably will reject the null hypothesis because .001 is statistically significantly different than the p value. It depends. Each statistic test answers a different type of question. In order to make charts you need special software by IBM. It is very difficult to make these charts in excel. Wow..you really need stats and a masters level quant class to do this. I just completed PHD level Quant and it was hard. Glad to say I scored really high but had to work on it like ten hours a day ..no joke. Good luck.

nurse.sandi · Apr 30, 2011

If you could find an article where it is already interpreted and then cite, if this meets the criteria of your assignment. That may work. I can not imagine why they think someone would know this without the prereqs to this type of assignment.

nurse.sandi · Apr 30, 2011

If you know how to work around excel. You could make bar charts. They are simple. If you sent me a pm. I might be able to help you.

nurse.sandi · Apr 30, 2011

I'm working on evaluating a pediatric randomized study. Our assignment was finding a topic from a nursing journal that obtains to maternal/child nursing, that contains a controlled, randomized study. I found several that contain simpler data, but I really liked the ideas behind the study on therapeutic play, because it speaks a lot to how we handle children as nurses.
Anyway, the data is not set up by percentages, but rather by Mean, SD, t-values, and p-values. I haven't ever taken a statistics class, so I don't really know the clinical significance of the final published results.
For example:
Childrens preoperative state
Experimental
Mean: 34.36
SD: 8.08
Control
Mean: 38.60
SD: 8.52
T-value: 3.6
P-value: .001
If anyone can give me some insight into how to better interpret the data, it'd be really appreciated. I found a couple sites that talked about t-test values, but still couldn't understand how to apply that to the results here.

OK the means are the average. The means are similiar but there is a small difference. The standard deviations are very close. The t value and the p value are what you need to interpret. The t value would be way higher than the alpha if it is set at .01 or .05. The idea is to disprove the theory. You want to either reject or retain the null hypothesis. So, what is the hypothesis you are trying to disprove? :idea:

delilas · Apr 30, 2011

I just finished a 7000 level Quantitative methods and analysis class. We need an alpha level. What is that set up as? Most of the time the alpha is set at .05. With a p=.001, you probably will reject the null hypothesis because .001 is statistically significantly different than the p value.

There is no alpha set in the paper...I even ctrl+F to search it and got nothing, so perhaps it is set at the 0.5? And haha, now I'm scared. If this is Master's level, perhaps I really should look for a paper that has data reported a different way.

If you know how to work around excel. You could make bar charts. They are simple. If you sent me a pm. I might be able to help you.

Thanks for the offer! I know how to use it, but it would be easier to display with percentages or simpler statistics, since no one in a basic ADN course is going to really understand the mean and it's clinical significance in the data.

If you could find an article where it is already interpreted and then cite, if this meets the criteria of your assignment. That may work.

Not a bad idea, I'll have to dig around some more!

Whispera, MSN, RN · Apr 30, 2011

Controlled, randomized study just means the subjects were randomly put into experimental and control groups, and that there WERE a control and an experimental group.

You can make simple charts graphs showing before and after. The SD and mean values are the same pre- and post-? Doesn't that mean there was no change overall? Maybe your charts and graphs can show individual differences rather than composite?

If I was you, I'd consider working on the articles that had simpler statistics even though this one is more interesting to you. This doesn't seem do-able as you are thinking it has to be done.

nurse.sandi · Apr 30, 2011

You can use pie charts to show percentages.