“Contracting Out” Statistical Analysis of Dissertation

Discussion in 'Off-Topic Discussions' started by me again, Apr 11, 2003.

Loading...
  1. me again

    me again Well-Known Member

    A professor of mine said that he “contracted out” the statistical analysis portion of his doctoral dissertation. He collected all of the data and then the hired someone to do the mythological statistical computations. He said it is perfectly acceptable, as long as you “give credit” to the person who did the computations. He graduated from Nova in 2000.

    Thoughts? :confused: :)
     
  2. Rich Douglas

    Rich Douglas Well-Known Member

    Unless your school specifically prohibits it, you can hire outside consultants to assist with data collection, analyses, even editing. But the work must be yours, not theirs.

    The hardest part of using statistical methods is deciding which method to use. Only you can decided that, and you have to do it before executing your study, so you know what data to collect and how, along with which inferential statistical procedures you will employ and why. Then there's the data collection itself, which you'll have to do, or supervise.

    The actual analysis itself is a snap, as long as you use a computer program and you understand your statistical procedures.

    Of course, if you know your statistics, collect your data, and use a computer program, why do you need someone else? And if you don't, you might find it hard to explain your results to someone else if you don't understand the methodology (its strengths and its weaknesses as they apply to your study).

    Many people yearn for outside help because they don't understand quantitative methods. This is the absolute worst reason for doing it. You have to know it inside and out.

    For my study, I used an online survey, downloaded the data, ported it into my stats program, and went to work on the data. But the results took me in a couple of additional directions that I had not anticipated. Thankfully, I understood the statistics and could visualize the additional analyses I would perform. Also, when the data come out funny, you need to be able to recognize this and fix the problem--hard to do if you don't understand the methods and are relying on someone else.

    Then there's the defense. When you've hired someone to do your stats for you, and one of your committee members asks about family error in post-hoc testing (after an ANOVA), and you need to explain that the Tukey-t test already accounts for it, negating the need to make the Bonferroni ajustment, I hope you get it right. Because the "stats-boy" you hired won't be there to explain it for you.

    Other than that, go for it.
     
  3. roysavia

    roysavia New Member

    Some people are not good at statistical computations. Gathering data and analyzing it with chi squared tests or regression analysis models is not only time consuming but difficult.
    I consider myself to be fairly proficient at statistics, however I would agree with you. There is nothing wrong with contacting out the statistical analysis portion of a dissertation as long as you cite source of this information.

    There are several online dissertation assistance companies that provide this service. You can probably cut your research time by 1/3 by signing a contract with one of these companies. Be prepared to pay between $1,000 and $4,000+ for this service.
     
  4. Rich Douglas

    Rich Douglas Well-Known Member

    If you are going to defend your dissertation, you have to understand your conclusions.

    If you want to understand your conclusions, you have to understand the analyses performed on the data.

    If you want to understand the analyses performed on the data, you have to understand inferential statistical methods.

    If you understand inferential statistical methods, why would you need to contract it out?

    It is not as simple as giving credit. If you can't explain the analytical methods used to draw your conclusions, how can you or anyone else be confident in them?

    Inferential statistics is just that, "inferential." You draw some data and you infer things about the population(s). But you can draw incorrect inferences if you do not apply the methods correctly. Also, you have to be familiar with the methods' limitations and weaknesses. This is all part of being a quantitative researcher.

    Again, How can anyone be confident in a researcher's findings if that researcher doesn't understand them? :rolleyes:
     
  5. me again

    me again Well-Known Member

    Good points, but...

    Richard,

    You make some good points. However, it has been contracted out in the past and people have obtained their doctoral degrees using this method. :p

    That's my only point. :)

    I'm not even in a doctoral program -- yet. :eek:
     
  6. roysavia

    roysavia New Member

    Re: Good points, but...

    Rich does make some very good points. However, of all the PhD's awarded by RA universities during the past 20 years, a small percentage of those who graduated most likely contracted out sections of their dissertation to third party sources.
    I work with someone (name withheld) who earned a PhD in Economics from the University of Toronto. She was at the ABD stage in her program. Apparently she was having considerable difficulty with her data analysis and inferential statistical methods. She eventually sought help from an independent consulting company. Her advisory committee was aware of this and permitted her to use their services. She was able to complete her dissertation and successfully defend it.
    If I recall my conversation with her correctly, she spent some time with her consultant reviewing her existing methodology, modifying it and streamlining the analysis.
     
  7. cehi

    cehi New Member

    Rich has made some excellent points. I agree with most of the points, especially related to the fact how can there be confidence in the researcher findings if the researcher does not know them. This to me, create another preamble for the researcher to properly select a research topic that would allow a proper research question(s) to be established, and would eventually allow the researcher to assign a proper method(s) to analyze the available data to make conclusions and recommendations. ofcourse, there are so many scientific researches that justify the use of inferential statistical methods. There are however, many researches that are also qualitative in nature whereby descriptive statistical or narrative methods may be used to still come up with some reasonable conclusions from the analyzed data.

    Yes, again, Rich is right. You would save yourself lots of headaches by applying method(s) to analyze data that you can effectively defend. This is more of a researcher beware!!!!
     
  8. obecve

    obecve New Member

    With the software that exists, it seems to me a horrible waste of time and money to "hire out your statistics". A program such as SPSS is easily used and understood, works well with ACCESS or Excel and is wonderfuly fast. I completed a discriminant analysis, cluster analysis and more than 50 ANOVA's in a couple of days. My professor and I worked closely together to understand the results. If I was to defend it...I felt I neded to know it. Yes you can hire out statistics, but I don't think you need to, particularly if it is a matter of fear about stats. Docs are supposed to understand research and stats. Most recent software makes it easy and gives you ownership of the data.
     
  9. Rich Douglas

    Rich Douglas Well-Known Member

    Obviously, I agree with Dr. O'Brien. I used GB-STAT, which allowed me to import the survey data I collected online, then apply the desired statistical tests to it. The program also had graphics capabilities, some of which I used. I was also able to take the GB-STAT data and export it to Microsoft Excel, which allowed me to re-code some of it. Also, Excel's graphics are good, so I used it to create more charts.

    Having to manipulate the data myself brought me a lot closer to it. I got more confident in the results, I understood them to a finer degree, and I was able to take the results into unanticipated directions.

    When I first enrolled in Union, internet-based search services (like Dialog and BRS) weren't generally available. I remember paying $75 for a search of ERIC through a local university library. It was a one-time shot which gave me many citations, but I couldn't follow up without ponying up another bunch of bucks. Like statisticis handled by someone else, I was really detached from the search process. I couldn't really "surf" through the search, letting it take me in unanticipated directions. Well, now you can search ERIC and hundreds of other databases for free, online, at home, at your convenience. Literature reviews have evolved from "just enough" to "just right."

    Statistics are no different.
     
  10. obecve

    obecve New Member

    I strongly agree with Rich. The cool thing about doing the stats yourself is that you then own it better than anyone else. You are the expert!
     
  11. Guest

    Guest Guest

    And isn't this the objective of earning a doctorate?
     
  12. Dr. Edwards

    Dr. Edwards member

    Pardon my ignorance, but the notion of "contracting out the statistical analysis portion" of a thesis is certainly new to me.

    I do not expect all graduate students to do "proofs" on statistical equations, but I cannot imagine anyone in a Masters program not having the ability to understand the fundamentals of stats, and to apply them to the collected data.

    Computer programs are great, but you need to know the basics of stats first. As suggested, you need to know your math before you can defend your thesis competently.

    A suggestion: Consider taking a 1st year-level stats or math course and do some real work of your own.
     
  13. Dr PK

    Dr PK member

    Re: Re: “Contracting Out” Statistical Analysis of Dissertation

    I concur totally. This 'contracting out' raises serious doubts in my mind as to what extent the work could be argued to be that of Mr Douglas ... and there also veiled comments about it being a bit patched up as well, or rehashed, after results appeared from the 'contracted' analysis.

    Neither I nor any of my co-supervisors would agree to this arrangement, but then this is not only a non Ivy League US degree which raises concerns often outside the US as to real quality work that has been done, but one from Union which does not engender the highest accolades in the academic world anyway.

    However, as Mr Douglas teaches at an unaccredited school and is also a trainer in industry, this probably doesn't matter that much.

    Mr Douglas' much personally touted thesis begs to be read and analysed in detail ... I just might get some of my freinds together to do that. At least it will get us an idea of what Union accepts, as some of their graduates try for jobs out here at times. The last product I read from there was very 'average', but then it is nowhere near top ranking is it?
     
    Last edited by a moderator: Apr 15, 2003
  14. drwetsch

    drwetsch New Member

    Can only comment on my own experience but I did all the stats on my dissertation myself. To that end I learned alot and enjoyed using the computer based statistical tools available. The tools allowed me to better manage the data and easily create models. The thought never even came into my brain that I could contract out this work. I would wonder how a Ph.D. who did contract out is able to mentor doctoral students who are trying to do their own stats?

    John
     
  15. Dr. Edwards

    Dr. Edwards member

    That's a darn good question. We now have a case of the blind leading the blind.
     
  16. roysavia

    roysavia New Member

    As I mentioned in an earlier post, I know one person who contracted out her statistical analysis. She decided not to teach for a living.
     
  17. me again

    me again Well-Known Member

    I see no reason why not and I am a kind man: Freely granted. ;)
    It was new to me also. Additionally, the doctor was referring to his doctoral program, which he completed at Nova in 2000.
    Chief,

    You didn't read my post very carefully, did you? And you're a doctor? I cannot imagine a doctor not having the ability to understand the fundamentals of a simple post and then to apply it to a chat forum. :rolleyes:

    A suggestion: Consider taking a 1st year-level reading and comprehension course and do some real comprehension of your own.

    Please note that the professor was referring to his doctoral program; he was not referring to the Masters program that we are in. :rolleyes:

    I hope you don't "read into" your student's papers like you "read into" my post. :rolleyes:

    Now about that first-year reading & comprehension class...
     
  18. me again

    me again Well-Known Member

    Ah, that felt good...

    Several years ago while doing undergraduate work, I quickly learned to never argue with your professor, even if he is clearly wrong because that could adversely affect your grade. :eek:

    Fortunately, I’m no longer doing undergraduate work and this is just a chat forum where we can freely express our minds. :D

    We're all grown ups now. :)
     
  19. Guest

    Guest Guest

    Re: Ah, that felt good...

    Wrong, me again!

    I must deduct two letter grades from your post for being argumentative and insubordinate.

    C+

    ;)
     
  20. me again

    me again Well-Known Member

    Dr. Edwards,

    Please forgive my boorish manners, which are a rarity that are seldom seen. Regardless of the brash banter that may occur, I sincerely welcome you to this educational forum and I do hope you’ll stay. :)
     

Share This Page