Return-Path: <nifl-family@literacy.nifl.gov> Received: from literacy (localhost [127.0.0.1]) by literacy.nifl.gov (8.10.2/8.10.2) with SMTP id i0UERlI24884; Fri, 30 Jan 2004 09:27:47 -0500 (EST) Date: Fri, 30 Jan 2004 09:27:47 -0500 (EST) Message-Id: <61E158A7.1C543305.0A349A3F@aol.com> Errors-To: listowner@literacy.nifl.gov Reply-To: nifl-family@literacy.nifl.gov Originator: nifl-family@literacy.nifl.gov Sender: nifl-family@literacy.nifl.gov Precedence: bulk From: AWilder106@aol.com To: Multiple recipients of list <nifl-family@literacy.nifl.gov> Subject: [NIFL-FAMILY:1777] Re: NIFL, research X-Listprocessor-Version: 6.0c -- ListProcessor by Anastasios Kotsikonas Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=iso-8859-1 X-Mailer: Atlas Mailer 2.0 Status: O Content-Length: 2576 Lines: 32 Folks, I thought I might have sounded too jaundiced yesterday, so here is a brief outline of what must be covered in any scientifically-based research. 1) What's your question? What do you want to know? 2) How will you answer the question? What method(s) are you going to use? 3) How do you plan to counter any threats to validity? Answer any critiques about your method? (Ex: "researcher bias" could be countered by having an independent researcher monitor process and outcomes) 4) How will you analyse your data? 5) To what population do you plan to generalize your results? You can see that all these steps interlock, like a puzzle. For example, the question and the method influence the populations you can generalize to. First, whatever you do, you have to disprove the "null hypothesis." The null hypothesis says in effect that there is nothing here to study--for example, it doesn't matter how you teach adult students, the results are always the same. The thinking about the study starts when you say something like: Hmmm, there is a considerable body of literature (studies) that show that it does matter what teaching methods are used. How can I find out what works best? Then you write your question, like: Among 4 possible methods, what works best? You have to choose a sample, and the preference is for a random sample, of large size, so all the variables that might influence"best method" are canceled out--age, gender, ethnicity, where the class meets, time spent in class, and so on. The more random the sample, the larger the generalizability: "X method works best in all cases." This is an extremely simplified example, almost comically so, so anyone who takes it as the answer to scientifically-based research would be dead wrong. Please don't do that. It is very important to look at who is in the sample--for example, All nurses in the HRT studies? (I may be wrong on this sample.) College students are routinely asked to be sample members because researchers are usually linked to universities. And on and on. An "opportunistic sample" may be "anyone you can get." The back pages of the "Harvard University Gazette," a weekly, run want ads for sample members--"Anyone between 20 and 40 years for self-esteem study." Sometimes payment is offered. Does payment draw people in? Or is payment only reasonable compensation for time spent? These two questions have both ethical and methodological implications. Standardized tests must list who the test has been "normed" on--that is your sample. Andrea
This archive was generated by hypermail 2b30 : Thu Dec 23 2004 - 09:47:04 EST